{"meta":{"count":302,"source":"src/data"},"items":[{"id":"openai","slug":"openai","name":"OpenAI","website":"https://openai.com","tagline":"Default when you want maximum capability and ecosystem glue. Multi-home critical paths behind OpenAI-compat…","description":"Default when you want maximum capability and ecosystem glue. Multi-home critical paths behind OpenAI-compat so Anthropic/Gemini remain exit ramps. Strengths: Broadest API surface (tools, multimodal, embeddings, realtime); Best ecosystem/docs/SDK gravity; Fast model cadence and enterprise ZDR options.","categoryIds":["cat-inference-apis","cat-media-gen"],"jtbdIds":["frontier-chat","agent-tools","multimodal","embeddings","coding-agent","long-context-agents","web-grounded-qa","code-interpreter","voice-agents","synthetic-eval-data"],"regions":["global","us"],"pricingNotes":"GPT-5.6 tiers (Sol/Terra/Luna) roughly $5/$30 to $1/$6 per 1M tokens; prompt caching with explicit breakpoints; batch discounts. Media (images/audio) billed separately.","strengths":["Broadest API surface (tools, multimodal, embeddings, realtime)","Best ecosystem/docs/SDK gravity","Fast model cadence and enterprise ZDR options"],"weaknesses":["Premium pricing vs open-model hosts","Rate limits and capacity can bite at scale","Policy/safety refusals sometimes block agent flows"],"tags":["llm","frontier","multimodal","tools","embeddings"],"scores":{"capabilityFit":9.6,"pricePerformance":6.2,"latencyReliability":8.8,"developerDx":9.5,"privacyPosture":7.8,"lockInRisk":7.2,"maturity":9.8,"overall":8.22,"kind":"curated_estimate","confidence":"high","asOf":"2026-09-14","sources":["OpenAI GPT-5.6 announcement: https://openai.com/index/gpt-5-6/","OpenAI API: https://platform.openai.com"],"notes":"Default when you want maximum capability and ecosystem glue. Multi-home critical paths behind OpenAI-compat so Anthropic/Gemini remain exit ramps."},"lastReviewed":"2026-09-14","status":"active"},{"id":"anthropic","slug":"anthropic","name":"Anthropic","website":"https://www.anthropic.com","tagline":"Depot default for serious coding and production agents. Prefer Sonnet for cost/quality; reserve Opus for hard…","description":"Depot default for serious coding and production agents. Prefer Sonnet for cost/quality; reserve Opus for hard reasoning. Pair with Bedrock/Vertex if procurement demands cloud wrappers. Strengths: Best-in-class for coding and careful tool-using agents; Strong enterprise privacy/compliance narrative; Excellent long-horizon agent behavior.","categoryIds":["cat-inference-apis"],"jtbdIds":["frontier-chat","coding-agent","long-context-agents","agent-tools","enterprise-llm","eu-sovereign-llm","multimodal","code-interpreter","prompt-caching-optimization","diff-aware-pr-review"],"regions":["global","us"],"pricingNotes":"Sonnet 5 ~$2/$10 per 1M; Opus ~$5/$25; Haiku 4.5 ~$1/$5; Fable higher. Prompt caching + batch. Managed Agents session-hour fees. US-only inference ~1.1x.","strengths":["Best-in-class for coding and careful tool-using agents","Strong enterprise privacy/compliance narrative","Excellent long-horizon agent behavior"],"weaknesses":["Narrower multimodal/media than OpenAI/Google","Can be capacity-constrained on Opus","Slightly less SDK ecosystem breadth than OpenAI"],"tags":["llm","frontier","coding","agents","claude"],"scores":{"capabilityFit":9.7,"pricePerformance":6.8,"latencyReliability":8.7,"developerDx":9.2,"privacyPosture":9,"lockInRisk":6.5,"maturity":9.4,"overall":8.42,"kind":"curated_estimate","confidence":"high","asOf":"2026-10-19","sources":["Anthropic pricing: https://www.anthropic.com/pricing","Claude Sonnet 5: https://www.anthropic.com/news/claude-sonnet-5"],"notes":"Depot default for serious coding and production agents. Prefer Sonnet for cost/quality; reserve Opus for hard reasoning. Pair with Bedrock/Vertex if procurement demands cloud wrappers."},"lastReviewed":"2026-10-19","status":"active"},{"id":"google-ai","slug":"google-ai","name":"Google AI (Gemini)","website":"https://ai.google.dev","tagline":"Best value frontier stack and multimodal king. Use Flash for high-volume agent loops; Pro when quality…","description":"Best value frontier stack and multimodal king. Use Flash for high-volume agent loops; Pro when quality matters. Prefer Vertex for enterprise compliance. Strengths: Outstanding multimodal + long context; Aggressive price/performance on Flash; Tight GCP/Vertex enterprise path.","categoryIds":["cat-inference-apis","cat-media-gen"],"jtbdIds":["frontier-chat","multimodal","long-context-agents","cheap-volume-llm","agent-tools","embeddings"],"regions":["global"],"pricingNotes":"Gemini 3.x Pro often ~$2/$12 under 200K; Flash tiers much cheaper. 1M context. Vertex for enterprise. Batch half-price common.","strengths":["Outstanding multimodal + long context","Aggressive price/performance on Flash","Tight GCP/Vertex enterprise path"],"weaknesses":["API surface and versioning can feel fragmented (AI Studio vs Vertex)","Agent tool-calling polish historically behind Anthropic/OpenAI","Quota surprises on free/unpaid tiers"],"tags":["llm","gemini","multimodal","long-context","value"],"scores":{"capabilityFit":9.2,"pricePerformance":8.8,"latencyReliability":8.5,"developerDx":8.2,"privacyPosture":8,"lockInRisk":6,"maturity":9.6,"overall":8.43,"kind":"curated_estimate","confidence":"high","asOf":"2026-10-19","sources":["Google AI for Developers: https://ai.google.dev","Gemini API: https://ai.google.dev/gemini-api"],"notes":"Best value frontier stack and multimodal king. Use Flash for high-volume agent loops; Pro when quality matters. Prefer Vertex for enterprise compliance."},"lastReviewed":"2026-10-19","status":"active"},{"id":"azure-openai","slug":"azure-openai","name":"Azure OpenAI","website":"https://azure.microsoft.com/products/ai-services/openai-service","tagline":"Pick when IT already standardized on Azure and needs VNET/private endpoints. Otherwise go direct OpenAI and…","description":"Pick when IT already standardized on Azure and needs VNET/private endpoints. Otherwise go direct OpenAI and save the cognitive tax. Strengths: Enterprise identity, networking, and compliance story; Regional data residency; Fits existing Azure estates.","categoryIds":["cat-inference-apis"],"jtbdIds":["enterprise-llm","frontier-chat"],"regions":["global","us","eu"],"pricingNotes":"OpenAI models via Azure with regional deployment; enterprise agreements, PTUs, and Microsoft billing. Often premium vs direct OpenAI.","strengths":["Enterprise identity, networking, and compliance story","Regional data residency","Fits existing Azure estates"],"weaknesses":["Provisioning friction vs direct OpenAI","Model lag and SKU complexity","Higher effective cost and lock-in to Azure"],"tags":["llm","enterprise","azure","openai"],"scores":{"capabilityFit":9,"pricePerformance":5.5,"latencyReliability":9,"developerDx":7,"privacyPosture":8.8,"lockInRisk":7.8,"maturity":9.5,"overall":7.62,"kind":"curated_estimate","confidence":"high","asOf":"2026-09-14","sources":["Azure OpenAI Service: https://azure.microsoft.com/products/ai-services/openai-service"],"notes":"Pick when IT already standardized on Azure and needs VNET/private endpoints. Otherwise go direct OpenAI and save the cognitive tax."},"lastReviewed":"2026-09-14","status":"active"},{"id":"amazon-bedrock","slug":"amazon-bedrock","name":"Amazon Bedrock","website":"https://aws.amazon.com/bedrock/","tagline":"Enterprise multi-model switchboard. Great when AWS is mandatory; mediocre when you just want the best Claude…","description":"Enterprise multi-model switchboard. Great when AWS is mandatory; mediocre when you just want the best Claude experience. Strengths: One API across many foundation models; Deep AWS IAM/VPC/compliance integration; Strong enterprise procurement path.","categoryIds":["cat-inference-apis"],"jtbdIds":["enterprise-llm","multi-model-router","rag-enterprise","hipaa-baa-llm"],"regions":["global","us","eu","ap"],"pricingNotes":"Pay-per-token for Claude, Llama, Amazon, Cohere, etc.; Provisioned Throughput for steady load. AWS marketplace billing.","strengths":["One API across many foundation models","Deep AWS IAM/VPC/compliance integration","Strong enterprise procurement path"],"weaknesses":["DX and latency often worse than first-party APIs","Feature lag vs Anthropic/OpenAI native","AWS pricing opacity"],"tags":["llm","enterprise","aws","multi-model"],"scores":{"capabilityFit":8.8,"pricePerformance":6,"latencyReliability":9.2,"developerDx":6.8,"privacyPosture":9,"lockInRisk":7,"maturity":9.6,"overall":7.75,"kind":"curated_estimate","confidence":"high","asOf":"2026-09-14","sources":["Amazon Bedrock: https://aws.amazon.com/bedrock/"],"notes":"Enterprise multi-model switchboard. Great when AWS is mandatory; mediocre when you just want the best Claude experience."},"lastReviewed":"2026-09-14","status":"active"},{"id":"cohere","slug":"cohere","name":"Cohere","website":"https://cohere.com","tagline":"Use for retrieval stacks (embed+rerank), not as your primary agent brain. Quietly excellent at enterprise RAG.","description":"Use for retrieval stacks (embed+rerank), not as your primary agent brain. Quietly excellent at enterprise RAG. Strengths: Excellent rerank + embed for RAG; Enterprise/private deployment focus; Strong retrieval-oriented models.","categoryIds":["cat-inference-apis"],"jtbdIds":["embeddings","rag-enterprise","cheap-volume-llm","eu-sovereign-llm","rag-rerank"],"regions":["global","us","eu","ca"],"pricingNotes":"Command/Rerank/Embed families; competitive enterprise RAG pricing; private deployment options.","strengths":["Excellent rerank + embed for RAG","Enterprise/private deployment focus","Strong retrieval-oriented models"],"weaknesses":["Not a frontier general chat leader","Smaller ecosystem than OpenAI/Anthropic","Less agent-tooling mindshare"],"tags":["llm","embeddings","rerank","rag","enterprise"],"scores":{"capabilityFit":7.8,"pricePerformance":8,"latencyReliability":8.4,"developerDx":7.8,"privacyPosture":8.6,"lockInRisk":4.5,"maturity":8.8,"overall":7.95,"kind":"curated_estimate","confidence":"high","asOf":"2026-10-19","sources":["Cohere: https://cohere.com","Cohere docs: https://docs.cohere.com"],"notes":"Use for retrieval stacks (embed+rerank), not as your primary agent brain. Quietly excellent at enterprise RAG."},"lastReviewed":"2026-10-19","status":"active"},{"id":"mistral","slug":"mistral","name":"Mistral AI","website":"https://mistral.ai","tagline":"Default EU-friendly LLM vendor. Pair with Mistral-hosted or self-host open weights when sovereignty matters.","description":"Default EU-friendly LLM vendor. Pair with Mistral-hosted or self-host open weights when sovereignty matters. Strengths: Strong EU/sovereignty story; Good open-weight releases; Solid coding mid-tier models.","categoryIds":["cat-inference-apis"],"jtbdIds":["cheap-volume-llm","eu-sovereign-llm","coding-agent"],"regions":["eu","global"],"pricingNotes":"La Plateforme API; mix of open and commercial models; EU-friendly positioning; competitive mid-tier pricing.","strengths":["Strong EU/sovereignty story","Good open-weight releases","Solid coding mid-tier models"],"weaknesses":["Frontier quality usually trails Anthropic/OpenAI","Smaller US enterprise footprint","Catalog churn"],"tags":["llm","eu","open-weights","coding"],"scores":{"capabilityFit":8.2,"pricePerformance":8.4,"latencyReliability":8,"developerDx":8,"privacyPosture":8.5,"lockInRisk":4,"maturity":8.2,"overall":8.05,"kind":"curated_estimate","confidence":"high","asOf":"2026-09-14","sources":["Mistral AI: https://mistral.ai","Mistral docs: https://docs.mistral.ai"],"notes":"Default EU-friendly LLM vendor. Pair with Mistral-hosted or self-host open weights when sovereignty matters."},"lastReviewed":"2026-09-14","status":"active"},{"id":"fireworks","slug":"fireworks","name":"Fireworks AI","website":"https://fireworks.ai","tagline":"Top pick for production open-model serving with real DX. Prefer over random GPU DIY when you want tokens not…","description":"Top pick for production open-model serving with real DX. Prefer over random GPU DIY when you want tokens not ops. Strengths: Fast production inference for open models; Strong fine-tune/post-train tooling; Good DX and model freshness.","categoryIds":["cat-inference-apis","cat-train-finetune"],"jtbdIds":["open-model-inference","cheap-volume-llm","finetune","ultra-low-latency"],"regions":["us","global"],"pricingNotes":"Serverless per-token + on-demand GPU; day-0 OSS model support; post-training stack. Free credits often available.","strengths":["Fast production inference for open models","Strong fine-tune/post-train tooling","Good DX and model freshness"],"weaknesses":["Not a frontier closed-model source","Unit cost can exceed pure discount hosts","Enterprise features still maturing vs hyperscalers"],"tags":["llm","open-models","inference","finetune","serverless"],"scores":{"capabilityFit":8.6,"pricePerformance":8.8,"latencyReliability":8.4,"developerDx":9,"privacyPosture":7.5,"lockInRisk":3.5,"maturity":8,"overall":8.35,"kind":"curated_estimate","confidence":"high","asOf":"2026-09-14","sources":["Fireworks AI: https://fireworks.ai","Best LLM API Providers (Fireworks blog): https://fireworks.ai/blog/best-llm-api-providers"],"notes":"Top pick for production open-model serving with real DX. Prefer over random GPU DIY when you want tokens not ops."},"lastReviewed":"2026-09-14","status":"active"},{"id":"together","slug":"together","name":"Together AI","website":"https://www.together.ai","tagline":"Best catalog breadth for OSS models. Watch minimums and lifetime spend gates before you design around a gated…","description":"Best catalog breadth for OSS models. Watch minimums and lifetime spend gates before you design around a gated model. Strengths: Huge open model catalog; Fine-tuning + inference under one roof; Competitive open-model pricing.","categoryIds":["cat-inference-apis","cat-train-finetune","cat-gpu-cloud"],"jtbdIds":["open-model-inference","finetune","cheap-volume-llm","multi-model-router"],"regions":["us","global"],"pricingNotes":"As of 2026-09-14: Open-model inference + fine-tune on one OpenAI-compatible API; competitive per-token rates (often ~$0.15–$0.90/MTok class depending on model). Strong for Llama/Qwen/DeepSeek families; Dedicated endpoints available.","strengths":["Huge open model catalog","Fine-tuning + inference under one roof","Competitive open-model pricing"],"weaknesses":["Billing tiers/spend gates can surprise mid-build","Quality variance across catalog models","Support uneven at small spend"],"tags":["llm","open-models","finetune","gpu"],"scores":{"capabilityFit":8.5,"pricePerformance":9,"latencyReliability":8,"developerDx":8.4,"privacyPosture":7.2,"lockInRisk":3.5,"maturity":8.2,"overall":8.2,"kind":"curated_estimate","confidence":"high","asOf":"2026-09-14","sources":["Together AI: https://www.together.ai","Together pricing: https://www.together.ai/pricing","Together docs: https://docs.together.ai"],"notes":"Best catalog breadth for OSS models. Watch minimums and lifetime spend gates before you design around a gated model."},"lastReviewed":"2026-09-14","status":"active"},{"id":"groq","slug":"groq","name":"Groq","website":"https://groq.com","tagline":"Use when latency is the product. Do not build your whole company on one Groq model SKU — they rotate.","description":"Use when latency is the product. Do not build your whole company on one Groq model SKU — they rotate. Strengths: Blazing tokens/sec latency; Simple API and free tier for prototypes; Great for real-time agent loops on supported models.","categoryIds":["cat-inference-apis"],"jtbdIds":["ultra-low-latency","cheap-volume-llm","open-model-inference","speech-to-text","content-moderation","llm-input-guardrails","prompt-injection-firewall"],"regions":["us","global"],"pricingNotes":"Per-token with free tier; LPU inference; narrow production model set (Llama/Qwen/GPT-OSS etc.); batch discounts. Llama Guard 4 12B (model id meta-llama/llama-guard-4-12b) is still documented at console.groq.com/docs/model/meta-llama/llama-guard-4-12b as an open safety classifier; its per-token price figure was removed because it couldn't be re-sourced on 2026-10-21 (no price on that page or on console.groq.com/docs/models as of 2026-10-21). See the stored rows for sourced Groq prices, including the Llama Prompt Guard 2 classifiers.","strengths":["Blazing tokens/sec latency","Simple API and free tier for prototypes","Great for real-time agent loops on supported models"],"weaknesses":["Limited model variety","Aggressive model deprecations","Not for frontier closed models"],"tags":["llm","lpu","latency","open-models","llama-guard","moderation","safety"],"scores":{"capabilityFit":7.8,"pricePerformance":9,"latencyReliability":8.2,"developerDx":8.8,"privacyPosture":7,"lockInRisk":5,"maturity":7.8,"overall":7.95,"kind":"curated_estimate","confidence":"high","asOf":"2026-09-14","sources":["Groq: https://groq.com","GroqCloud: https://console.groq.com","Groq Llama Guard 4 docs: https://console.groq.com/docs/model/meta-llama/llama-guard-4-12b"],"notes":"Use when latency is the product. Do not build your whole company on one Groq model SKU — they rotate."},"lastReviewed":"2026-09-14","status":"active"},{"id":"cerebras","slug":"cerebras","name":"Cerebras","website":"https://www.cerebras.ai","tagline":"Best ultra-throughput play when your model is on Cerebras (official token matrix now pinned). Pair with…","description":"Best ultra-throughput play when your model is on Cerebras (official token matrix now pinned). Pair with Groq/Fireworks fallback for catalog holes and rate-limit spikes. Strengths: Extreme throughput on supported models; Differentiated hardware story; Attractive for high-QPS serving.","categoryIds":["cat-inference-apis","cat-gpu-cloud"],"jtbdIds":["ultra-low-latency","open-model-inference","cheap-volume-llm"],"regions":["us"],"pricingNotes":"As of 2026-09-14: Official public models API (api.cerebras.ai/public/v1/models, HuggingFace format) pins GPT-OSS-120B $0.35/$0.75 per 1M in/out (131k ctx) and Qwen3.8-27B $0.99/$1.49 (65k ctx). Marketing cerebras.ai/inference still shows $5 free credit; Developer PayGo from $10 + 10× rate limits; Enterprise dedicated. Official docs.cerebras.ai public-models endpoint is the token matrix of record — not PricePerToken.","strengths":["Extreme throughput on supported models","Differentiated hardware story","Attractive for high-QPS serving"],"weaknesses":["Narrow model/support surface vs GPU hosts","Availability and waitlists historically","Less commodity than NVIDIA ecosystem"],"tags":["llm","wafer-scale","latency","inference"],"scores":{"capabilityFit":8,"pricePerformance":8.2,"latencyReliability":7.8,"developerDx":7.5,"privacyPosture":7.2,"lockInRisk":5.5,"maturity":7.5,"overall":7.56,"kind":"curated_estimate","confidence":"high","asOf":"2026-09-14","sources":["Cerebras public models API (pricing): https://inference-docs.cerebras.ai/api-reference/models/public-models","Cerebras Inference: https://www.cerebras.ai/inference","Cerebras Inference API: https://inference.cerebras.ai"],"notes":"Best ultra-throughput play when your model is on Cerebras (official token matrix now pinned). Pair with Groq/Fireworks fallback for catalog holes and rate-limit spikes."},"lastReviewed":"2026-09-14","status":"active"},{"id":"sambanova","slug":"sambanova","name":"SambaNova","website":"https://sambanova.ai","tagline":"Public token matrix now exists — gpt-oss-120b $0.22/$0.59 is the value SKU. Indie agents still default…","description":"Public token matrix now exists — gpt-oss-120b $0.22/$0.59 is the value SKU. Indie agents still default Fireworks/Together/Groq unless SambaNova is already in-procurement. Strengths: High-performance enterprise inference; Custom silicon efficiency narrative; Enterprise sales motion.","categoryIds":["cat-inference-apis"],"jtbdIds":["ultra-low-latency","enterprise-llm","open-model-inference"],"regions":["us"],"pricingNotes":"As of 2026-09-14 (cloud.sambanova.ai/plans/pricing): MiniMax-M2.7 $0.60/$2.40 per 1M (cached in $0.06); DeepSeek V3.1/V3.2 $3/$4.50; gemma-4-31B-it $0.38/$1.15; gpt-oss-120b $0.22/$0.59; Llama 3.3 70B $0.60/$1.20; MiniMax-M3 $0.60/$2.40. RDU silicon, OpenAI-compat. Enterprise plans still sales-assisted.","strengths":["High-performance enterprise inference","Custom silicon efficiency narrative","Enterprise sales motion"],"weaknesses":["Self-serve DX weaker than Groq/Fireworks","Less transparent public pricing","Smaller indie community"],"tags":["llm","enterprise","custom-silicon"],"scores":{"capabilityFit":8,"pricePerformance":7.8,"latencyReliability":8.2,"developerDx":6.8,"privacyPosture":8,"lockInRisk":6,"maturity":8,"overall":7.53,"kind":"curated_estimate","confidence":"high","asOf":"2026-09-14","sources":["SambaNova Cloud pricing: https://cloud.sambanova.ai/plans/pricing","SambaNova Cloud plans: https://cloud.sambanova.ai/plans","SambaNova: https://sambanova.ai"],"notes":"Public token matrix now exists — gpt-oss-120b $0.22/$0.59 is the value SKU. Indie agents still default Fireworks/Together/Groq unless SambaNova is already in-procurement."},"lastReviewed":"2026-09-14","status":"active"},{"id":"deepinfra","slug":"deepinfra","name":"DeepInfra","website":"https://deepinfra.com","tagline":"Budget open-model workhorse and one of the cheapest large-catalog hosts. Great for evals and high-volume…","description":"Budget open-model workhorse and one of the cheapest large-catalog hosts. Great for evals and high-volume non-critical traffic; dual-run critical paths on Fireworks/Together. Strengths: Aggressive pricing; OpenAI-compat DX; Wide OSS model coverage.","categoryIds":["cat-inference-apis","cat-gpu-cloud"],"jtbdIds":["open-model-inference","cheap-volume-llm","budget-open-inference"],"regions":["us","global"],"pricingNotes":"As of 2026-09-14: Broad OpenAI-compatible catalog with model-specific token rates (e.g. Llama-3.3-70B Turbo $0.10/$0.32 per 1M in/out; DeepSeek-V4-Flash input $0.09, cached input $0.018 and output $0.18 per 1M (deepinfra.com/pricing as of 2026-10-21; not stored as rows); embeddings from $0.005/MTok). Dedicated GPU rentals (deepinfra.com/pricing, re-checked 2026-10-20): A100 80GB $1.49/GPU-hr, H100 $2.90, H200 $3.49, B200 $4.89, B300 $5.89. Standard/Priority(1.5x)/Flex(0.8x) tiers. No long-term contracts.","strengths":["Aggressive pricing","OpenAI-compat DX","Wide OSS model coverage"],"weaknesses":["Reliability/support less proven than Fireworks/Together at scale","Fewer enterprise compliance bells","Brand trust lower for regulated workloads"],"tags":["llm","open-models","budget","openai-compat"],"scores":{"capabilityFit":8,"pricePerformance":9.4,"latencyReliability":7.5,"developerDx":8.5,"privacyPosture":6.5,"lockInRisk":3,"maturity":7.2,"overall":7.95,"kind":"curated_estimate","confidence":"high","asOf":"2026-09-14","sources":["DeepInfra Pricing: https://deepinfra.com/pricing","DeepInfra: https://deepinfra.com","Cheapest inference providers Jul 2026: https://infrabase.ai/blog/cheapest-inference-providers"],"notes":"Budget open-model workhorse and one of the cheapest large-catalog hosts. Great for evals and high-volume non-critical traffic; dual-run critical paths on Fireworks/Together."},"lastReviewed":"2026-09-14","status":"active"},{"id":"openrouter","slug":"openrouter","name":"OpenRouter","website":"https://openrouter.ai","tagline":"Best multi-model router for agents that need failover and experimentation. Pin providers in production; don’t…","description":"Best multi-model router for agents that need failover and experimentation. Pin providers in production; don’t treat OpenRouter as your only SLA. Strengths: One key for dozens of models/providers; Fantastic for prototyping and failover; Transparent model playground.","categoryIds":["cat-inference-apis"],"jtbdIds":["multi-model-router","open-model-inference","frontier-chat","agent-tools","cost-monitoring"],"regions":["global"],"pricingNotes":"As of 2026-09-14: Multi-provider LLM router with OpenAI-compatible API; pass-through model pricing plus small platform fee. Best for prototyping multi-model and failover; production should pin critical paths to direct providers.","strengths":["One key for dozens of models/providers","Fantastic for prototyping and failover","Transparent model playground"],"weaknesses":["Extra platform fee and abstraction layer","End reliability = weakest upstream","Not ideal as sole production critical path without pinning providers"],"tags":["llm","router","multi-provider","openai-compat"],"scores":{"capabilityFit":8.8,"pricePerformance":7.8,"latencyReliability":7.6,"developerDx":9.2,"privacyPosture":6.8,"lockInRisk":4,"maturity":7.8,"overall":8,"kind":"curated_estimate","confidence":"high","asOf":"2026-09-14","sources":["OpenRouter: https://openrouter.ai","OpenRouter models: https://openrouter.ai/models","OpenRouter docs: https://openrouter.ai/docs"],"notes":"Best multi-model router for agents that need failover and experimentation. Pin providers in production; don’t treat OpenRouter as your only SLA."},"lastReviewed":"2026-09-14","status":"active"},{"id":"huggingface-inference","slug":"huggingface-inference","name":"Hugging Face Inference","website":"https://endpoints.huggingface.co","tagline":"Perfect for Hub-native workflows and custom models. For hard production latency, graduate to…","description":"Perfect for Hub-native workflows and custom models. For hard production latency, graduate to Fireworks/Together/Groq or your own GPUs. Strengths: Unmatched model Hub gravity; Easy path from Hub model to endpoint; EU options and open ecosystem.","categoryIds":["cat-inference-apis","cat-train-finetune"],"jtbdIds":["open-model-inference","finetune","eu-sovereign-llm","embeddings"],"regions":["global","us","eu"],"pricingNotes":"Inference Providers (pay-as-you-go access to partner inference providers plus Hugging Face's own hf-inference provider) + dedicated Inference Endpoints ('Pay for what you use, per minute' per https://endpoints.huggingface.co, fetched 2026-10-25; instance rates are the price rows: single-GPU x1 rows per GPU-hour, multi-GPU x2/x4/x8 rows per instance-hour (usd_per_instance_hour since v0.12.138, re-checked 2026-10-27 on https://huggingface.co/docs/inference-endpoints/pricing), never divided per GPU). The 'Serverless Inference API' name is retired: https://huggingface.co/docs/inference-providers/pricing (fetched 2026-10-25) says hf-inference 'used to be called “Inference API (serverless)” prior to Inference Providers' and 'just pay for what you use, with no markup from Hugging Face.' Monthly credits: 'Free Users None - yes (credits purchase required)'; PRO users and Team/Enterprise organizations get included monthly credits. Hub-centric. No price changed here; any price change goes through the reviewed --catalog step.","strengths":["Unmatched model Hub gravity","Easy path from Hub model to endpoint","EU options and open ecosystem"],"weaknesses":["Free accounts get no included Inference Providers credits ('Free Users None', credits purchase required, per https://huggingface.co/docs/inference-providers/pricing fetched 2026-10-25)","Not the fastest/cheapest dedicated inference","Ops complexity at scale vs Fireworks"],"tags":["llm","hub","open-models","endpoints"],"scores":{"capabilityFit":8.4,"pricePerformance":7.2,"latencyReliability":7.4,"developerDx":8.6,"privacyPosture":7.8,"lockInRisk":3.5,"maturity":9.2,"overall":7.95,"kind":"curated_estimate","confidence":"high","asOf":"2026-09-14","sources":["Hugging Face Inference Endpoints: https://huggingface.co/inference-endpoints","HF Hub: https://huggingface.co"],"notes":"Perfect for Hub-native workflows and custom models. For hard production latency, graduate to Fireworks/Together/Groq or your own GPUs."},"lastReviewed":"2026-09-14","status":"active"},{"id":"perplexity","slug":"perplexity","name":"Perplexity","website":"https://www.perplexity.ai","tagline":"Use when you want grounded answers without building search+cite yourself. For full agent stacks, prefer…","description":"Use when you want grounded answers without building search+cite yourself. For full agent stacks, prefer Exa/Tavily + your LLM. Strengths: Built-in web grounding; Strong answer-with-citations UX; Simple API for research agents.","categoryIds":["cat-inference-apis","cat-search-scrape"],"jtbdIds":["web-grounded-qa","search-augmented-llm"],"regions":["global"],"pricingNotes":"As of 2026-09-14 (docs.perplexity.ai pricing): Sonar $1/$1 per 1M in/out; Sonar Pro $3/$15; Sonar Reasoning Pro $2/$8; Sonar Deep Research $2/$8 + citation $2/1M + search queries $5/1K + reasoning $3/1M. Request fees per 1K (low/med/high context): Sonar $5/$8/$12; Sonar Pro & Reasoning Pro $6/$10/$14. Pro Search (Sonar Pro) $14/$18/$22. Search API $5/1K requests. Agent tools: web_search $0.0025, fetch_url $0.0005, people/finance_search $0.005, sandbox $0.03/session. Embeddings pplx-embed-v1-0.6b $0.004/1M; 4b $0.03/1M. Consumer Pro ≠ API billing. Note: Sonar Chat Completions→Agent API migration; Sonar supported until 2026-09-27.","strengths":["Built-in web grounding","Strong answer-with-citations UX","Simple API for research agents"],"weaknesses":["Less control than DIY search+LLM","Not a general frontier agent platform","Pricing/limits can shift with consumer product"],"tags":["llm","search","grounding","sonar"],"scores":{"capabilityFit":8.2,"pricePerformance":7.5,"latencyReliability":8,"developerDx":8,"privacyPosture":7,"lockInRisk":5.5,"maturity":8.5,"overall":7.67,"kind":"curated_estimate","confidence":"high","asOf":"2026-09-14","sources":["Perplexity API pricing: https://docs.perplexity.ai/docs/getting-started/pricing","Perplexity: https://www.perplexity.ai","Perplexity API docs: https://docs.perplexity.ai","Perplexity Sonar: https://docs.perplexity.ai/guides/sonar"],"notes":"Use when you want grounded answers without building search+cite yourself. For full agent stacks, prefer Exa/Tavily + your LLM."},"lastReviewed":"2026-09-14","status":"active"},{"id":"xai","slug":"xai","name":"xAI","website":"https://x.ai","tagline":"Durable public rate card — worth A/B vs OpenAI/Anthropic on general chat. X Search now bills per…","description":"Durable public rate card — worth A/B vs OpenAI/Anthropic on general chat. X Search now bills per posts/profiles (live since 2026-09-21 12:00 PT) — model cost against agent search peers carefully. Not the boring default for regulated agents. Strengths: Competitive frontier models; Growing API surface; Distinctive realtime/X-adjacent positioning.","categoryIds":["cat-inference-apis"],"jtbdIds":["frontier-chat","agent-tools","web-grounded-qa"],"regions":["us","global"],"pricingNotes":"As of 2026-09-29 (docs.x.ai/developers/pricing live): grok-4.7/4.6 $2/$6 per 1M in/out (<200k prompt; $4/$12 ≥200k); grok-4.5 same band; grok-4.3 / grok-4.20 $1.25/$2.50 (1M ctx; doubles ≥200k); grok-build-0.1 per 1M tokens under 200k prompt: input $1.00, cached input $0.20, output $2.00; at or above 200k: input $2.00, cached input $0.40, output $4.00 (docs.x.ai/developers/pricing as of 2026-10-21; not stored as rows). Cached input $0.20–$0.50. Tools: web_search $5/1k calls. **X Search still $5/1k posts + $10/1k profiles** (live since 2026-09-21 12:00 PT; parent/quoted posts count; every profile returned counts). Code exec $5/1k; collections_search $2.50/1k; **attachment_search $5/1k** (was $10/1k as of 2026-09-22 note). Batch 20% off selected 4.3/4.20 SKUs. Priority 2×. Imagine image $0.02–$0.05; video $0.05–$0.08/sec. Voice STS $0.08/min; STT $0.10/hr REST / $0.20/hr streaming; TTS $15/1M chars. File storage $0.025/GiB/day; collection storage $0.10/GiB/day; downloads $0.20/GiB.","strengths":["Competitive frontier models","Growing API surface","Distinctive realtime/X-adjacent positioning"],"weaknesses":["Ecosystem and enterprise maturity behind OpenAI/Anthropic","Policy/brand risk for some buyers","Docs/SDK depth still catching up"],"tags":["llm","grok","frontier"],"scores":{"capabilityFit":8.8,"pricePerformance":8,"latencyReliability":7.8,"developerDx":7.8,"privacyPosture":6.8,"lockInRisk":5.5,"maturity":7,"overall":7.68,"kind":"curated_estimate","confidence":"high","asOf":"2026-09-14","sources":["xAI API pricing: https://docs.x.ai/developers/pricing","xAI API docs: https://docs.x.ai","xAI console: https://console.x.ai"],"notes":"Durable public rate card — worth A/B vs OpenAI/Anthropic on general chat. X Search now bills per posts/profiles (live since 2026-09-21 12:00 PT) — model cost against agent search peers carefully. Not the boring default for regulated agents."},"lastReviewed":"2026-09-14","status":"active"},{"id":"ai21","slug":"ai21","name":"AI21 Labs","website":"https://www.ai21.com","tagline":"Enterprise RAG niche player. Only pick if you already standardized on AI21 or need their specific stack.","description":"Enterprise RAG niche player. Only pick if you already standardized on AI21 or need their specific stack. Strengths: Enterprise NLP/RAG heritage; Long-context Jamba models; Business-oriented tooling.","categoryIds":["cat-inference-apis"],"jtbdIds":["enterprise-llm","rag-enterprise","long-context-agents"],"regions":["global","us","il"],"pricingNotes":"As of 2026-09-14: Jamba family + enterprise LLM/RAG APIs; contextual answers products; pricing often enterprise/sales-led with usage tiers. Not a price leader vs open-model hosts; evaluate only if Jamba quality or AI21 stack already standardized.","strengths":["Enterprise NLP/RAG heritage","Long-context Jamba models","Business-oriented tooling"],"weaknesses":["Mindshare far behind frontier trio","Smaller agent ecosystem","Less indie DX buzz"],"tags":["llm","enterprise","jamba","rag"],"scores":{"capabilityFit":7.6,"pricePerformance":7.2,"latencyReliability":8.2,"developerDx":7.2,"privacyPosture":8,"lockInRisk":5,"maturity":8.4,"overall":7.5,"kind":"curated_estimate","confidence":"medium","asOf":"2026-09-14","sources":["AI21: https://www.ai21.com","AI21 Studio: https://studio.ai21.com","AI21 docs: https://docs.ai21.com"],"notes":"Enterprise RAG niche player. Only pick if you already standardized on AI21 or need their specific stack."},"lastReviewed":"2026-09-14","status":"active"},{"id":"deepseek","slug":"deepseek","name":"DeepSeek","website":"https://www.deepseek.com","tagline":"Unbeatable $/quality for many workloads — if your compliance team allows it. Otherwise host similar open…","description":"Unbeatable $/quality for many workloads — if your compliance team allows it. Otherwise host similar open weights on Fireworks/Together/DeepInfra. Strengths: Extreme price/performance; Strong coding/reasoning models; API broadly available.","categoryIds":["cat-inference-apis"],"jtbdIds":["cheap-volume-llm","coding-agent","open-model-inference"],"regions":["cn","global"],"pricingNotes":"As of 2026-09-14: Official API at api-docs.deepseek.com with very aggressive Flash/Pro token pricing (often cents per MTok on Flash tiers). Strong coding/reasoning open releases also hostable elsewhere. Geopolitical/compliance caveats for some orgs — host weights on Fireworks/Together/DeepInfra if direct API is blocked.","strengths":["Extreme price/performance","Strong coding/reasoning models","API broadly available"],"weaknesses":["Data residency/compliance concerns for many Western enterprises","Support/SLA expectations differ","Policy risk in regulated industries"],"tags":["llm","budget","coding","china"],"scores":{"capabilityFit":8.8,"pricePerformance":9.8,"latencyReliability":7.8,"developerDx":8,"privacyPosture":4.5,"lockInRisk":4,"maturity":7.5,"overall":7.95,"kind":"curated_estimate","confidence":"high","asOf":"2026-09-14","sources":["DeepSeek: https://www.deepseek.com","DeepSeek API docs: https://api-docs.deepseek.com","DeepSeek platform: https://platform.deepseek.com"],"notes":"Unbeatable $/quality for many workloads — if your compliance team allows it. Otherwise host similar open weights on Fireworks/Together/DeepInfra."},"lastReviewed":"2026-09-14","status":"active"},{"id":"vertex-ai","slug":"vertex-ai","name":"Google Vertex AI","website":"https://cloud.google.com/vertex-ai","tagline":"Enterprise Gemini path. Use AI Studio for prototypes, Vertex when security review starts asking questions.","description":"Enterprise Gemini path. Use AI Studio for prototypes, Vertex when security review starts asking questions. Strengths: Full ML platform + Gemini; GCP compliance and IAM; Batch and enterprise controls.","categoryIds":["cat-inference-apis","cat-train-finetune"],"jtbdIds":["enterprise-llm","multimodal","finetune","multi-model-router"],"regions":["global"],"pricingNotes":"As of 2026-09-14: Google Cloud Vertex AI model garden + Gemini; billed in GCP currency with enterprise controls, VPC-SC, and committed use discounts. Prefer Google AI Studio/API for indie DX; Vertex for enterprise residency/IAM.","strengths":["Full ML platform + Gemini","GCP compliance and IAM","Batch and enterprise controls"],"weaknesses":["Heavier DX than AI Studio","GCP lock-in","Cost complexity"],"tags":["llm","gcp","enterprise","gemini"],"scores":{"capabilityFit":9,"pricePerformance":7,"latencyReliability":9,"developerDx":7.2,"privacyPosture":8.8,"lockInRisk":7.5,"maturity":9.6,"overall":7.96,"kind":"curated_estimate","confidence":"high","asOf":"2026-09-14","sources":["Vertex AI: https://cloud.google.com/vertex-ai","Vertex AI pricing: https://cloud.google.com/vertex-ai/pricing"],"notes":"Enterprise Gemini path. Use AI Studio for prototypes, Vertex when security review starts asking questions."},"lastReviewed":"2026-09-14","status":"active"},{"id":"coreweave","slug":"coreweave","name":"CoreWeave","website":"https://www.coreweave.com","tagline":"Enterprise training/inference fleet pick. Skip if you just need a single H100 for an afternoon — use RunPod.","description":"Enterprise training/inference fleet pick. Skip if you just need a single H100 for an afternoon — use RunPod. Strengths: Enterprise K8s GPU cloud; Strong interconnect/training story; SOC2 and dedicated capacity.","categoryIds":["cat-gpu-cloud"],"jtbdIds":["multi-node-training","enterprise-gpu","production-inference-fleet","predictable-gpu","neocloud-gpu"],"regions":["us","eu"],"pricingNotes":"On-demand H100 ~$6+/hr class historically; B200 higher; Kubernetes-native (CKS); often sales for large capacity. Public company.","strengths":["Enterprise K8s GPU cloud","Strong interconnect/training story","SOC2 and dedicated capacity"],"weaknesses":["Expensive vs RunPod/Vast","Sales friction for some tiers","Overkill for single-GPU agents"],"tags":["gpu","kubernetes","enterprise","training"],"scores":{"capabilityFit":9.2,"pricePerformance":5.5,"latencyReliability":9,"developerDx":7.5,"privacyPosture":8.2,"lockInRisk":6.5,"maturity":8.8,"overall":7.71,"kind":"curated_estimate","confidence":"high","asOf":"2026-10-19","sources":["CoreWeave: https://www.coreweave.com"],"notes":"Enterprise training/inference fleet pick. Skip if you just need a single H100 for an afternoon — use RunPod."},"lastReviewed":"2026-10-19","status":"active"},{"id":"lambda","slug":"lambda","name":"Lambda","website":"https://lambda.ai","tagline":"Best boring GPU cloud for research labs who want clusters without CoreWeave enterprise theater.","description":"Best boring GPU cloud for research labs who want clusters without CoreWeave enterprise theater. Strengths: Clean ML-oriented UX; Predictable dedicated capacity; Strong cluster story for labs.","categoryIds":["cat-gpu-cloud","cat-train-finetune"],"jtbdIds":["multi-node-training","finetune","predictable-gpu","neocloud-gpu"],"regions":["us"],"pricingNotes":"On-demand H100 SXM ~$3.99–4.29/hr; 1-Click Clusters for multi-week reservations; no classic spot tier.","strengths":["Clean ML-oriented UX","Predictable dedicated capacity","Strong cluster story for labs"],"weaknesses":["Pricier than marketplace/neocloud budget tiers","Less interruptible discounting","US-centric"],"tags":["gpu","training","clusters","research"],"scores":{"capabilityFit":8.8,"pricePerformance":6.8,"latencyReliability":8.8,"developerDx":8.5,"privacyPosture":7.8,"lockInRisk":5.5,"maturity":8.6,"overall":7.97,"kind":"curated_estimate","confidence":"high","asOf":"2026-09-14","sources":["Lambda: https://lambda.ai"],"notes":"Best boring GPU cloud for research labs who want clusters without CoreWeave enterprise theater."},"lastReviewed":"2026-09-14","status":"active"},{"id":"runpod","slug":"runpod","name":"RunPod","website":"https://www.runpod.io","tagline":"Depot default for self-serve GPUs. Secure Cloud for anything customer-facing; Community/spot for batch.","description":"Depot default for self-serve GPUs. Secure Cloud for anything customer-facing; Community/spot for batch. Strengths: Best flexible middle ground for indie→prod; Serverless + pods + clusters; Banana.dev spiritual successor for many.","categoryIds":["cat-gpu-cloud"],"jtbdIds":["serverless-gpu","cheap-gpu","finetune","production-inference-fleet","predictable-gpu","batch-gpu"],"regions":["global","us","eu"],"pricingNotes":"Secure Cloud + Community Cloud; serverless per-second; H100 often ~$2–3+/hr depending tier; spot cheaper.","strengths":["Best flexible middle ground for indie→prod","Serverless + pods + clusters","Banana.dev spiritual successor for many"],"weaknesses":["Community Cloud reliability varies","Support quality uneven at low spend","Need discipline on storage/network costs"],"tags":["gpu","serverless","pods","budget"],"scores":{"capabilityFit":8.8,"pricePerformance":9,"latencyReliability":8,"developerDx":8.8,"privacyPosture":7,"lockInRisk":3.5,"maturity":8.2,"overall":8.32,"kind":"curated_estimate","confidence":"high","asOf":"2026-10-19","sources":["RunPod: https://www.runpod.io","GPU cloud comparisons: https://www.runpod.io/articles/alternatives/best-gpu-cloud-providers-for-ai-and-ml"],"notes":"Depot default for self-serve GPUs. Secure Cloud for anything customer-facing; Community/spot for batch."},"lastReviewed":"2026-10-19","status":"active"},{"id":"vast-ai","slug":"vast-ai","name":"Vast.ai","website":"https://vast.ai","tagline":"Cheapest GPUs if your job can die and restart. Never put customer PII training here without eyes wide open.","description":"Cheapest GPUs if your job can die and restart. Never put customer PII training here without eyes wide open. Strengths: Lowest sticker GPU prices; Huge heterogeneous inventory; Good for fault-tolerant batch.","categoryIds":["cat-gpu-cloud"],"jtbdIds":["cheap-gpu","batch-gpu","distributed-cheap-gpu"],"regions":["global"],"pricingNotes":"As of 2026-09-14: P2P GPU marketplace; H100 commonly ~$1.5–2.5/hr class depending on host/interruptibility; prices fluctuate by supply. Interruptible instances cheapest; verify host reliability and security posture. Never default for regulated PII training.","strengths":["Lowest sticker GPU prices","Huge heterogeneous inventory","Good for fault-tolerant batch"],"weaknesses":["No real SLA; host quality variance","Security/compliance unsuitable for many prod workloads","Ops overhead for reliability"],"tags":["gpu","marketplace","spot","budget"],"scores":{"capabilityFit":7,"pricePerformance":9.8,"latencyReliability":4.5,"developerDx":7.2,"privacyPosture":4,"lockInRisk":2.5,"maturity":7.5,"overall":6.94,"kind":"curated_estimate","confidence":"medium","asOf":"2026-09-14","sources":["Vast.ai: https://vast.ai","Vast.ai console: https://cloud.vast.ai","GPU price comparisons 2026: https://computeprices.com"],"notes":"Cheapest GPUs if your job can die and restart. Never put customer PII training here without eyes wide open."},"lastReviewed":"2026-09-14","status":"active"},{"id":"crusoe","slug":"crusoe","name":"Crusoe","website":"https://www.crusoe.ai","tagline":"Solid neocloud when you want SLA + energy story. Not the first stop for hackathon GPUs.","description":"Solid neocloud when you want SLA + energy story. Not the first stop for hackathon GPUs. Strengths: Sustainability + performance narrative; Dedicated infrastructure quality; Growing enterprise footprint.","categoryIds":["cat-gpu-cloud"],"jtbdIds":["enterprise-gpu","predictable-gpu","multi-node-training","low-carbon-gpu-inference"],"regions":["us"],"pricingNotes":"As of 2026-09-14: Energy-conscious neocloud GPU capacity; H100 typically mid-neocloud band with SLA-backed dedicated options; some large SKUs sales-gated. Strong sustainability narrative; DX less agent-native than RunPod/Modal.","strengths":["Sustainability + performance narrative","Dedicated infrastructure quality","Growing enterprise footprint"],"weaknesses":["Less indie self-serve mindshare than RunPod","Pricing not always cheapest","SKU availability varies"],"tags":["gpu","enterprise","sustainable"],"scores":{"capabilityFit":8.6,"pricePerformance":7.2,"latencyReliability":8.6,"developerDx":7.4,"privacyPosture":8,"lockInRisk":5.5,"maturity":8,"overall":7.76,"kind":"curated_estimate","confidence":"medium","asOf":"2026-09-14","sources":["Crusoe: https://www.crusoe.ai","Crusoe Cloud: https://www.crusoe.ai/cloud","Crusoe pricing context: https://computeprices.com"],"notes":"Solid neocloud when you want SLA + energy story. Not the first stop for hackathon GPUs."},"lastReviewed":"2026-09-14","status":"active"},{"id":"voltage-park","slug":"voltage-park","name":"Voltage Park","website":"https://www.voltagepark.com","tagline":"Now legitimately self-serve for H100 bursts at published $1.99–$2.49/hr. Prefer RunPod/Nebius for indie…","description":"Now legitimately self-serve for H100 bursts at published $1.99–$2.49/hr. Prefer RunPod/Nebius for indie experiments; Voltage Park when you want owned-fleet H100 clusters + clear list price without hyperscaler egress theater. Confirm Blackwell quotes separately. Strengths: Published on-demand H100 list prices ($1.99 / $2.49); Self-serve provisioning (~15 min) for bursts; Owned hardware narrative; no surprise ingress/egress/support fees claimed.","categoryIds":["cat-gpu-cloud"],"jtbdIds":["enterprise-gpu","multi-node-training"],"regions":["us"],"pricingNotes":"As of 2026-09-14 (official voltagepark.com/pricing + product pages): On-demand HGX H100 from $1.99/GPU-hr (Ethernet) and $2.49/GPU-hr (3200 Gbps InfiniBand); self-serve in ~15 min, no minimum term, marketed with no hidden ingress/egress/support fees. Dedicated Reserve = 6+ month contracts for 32–8000+ GPUs (sales). Merged with Lightning AI (2026) — Blackwell/GB SKUs still contact-sales.","strengths":["Published on-demand H100 list prices ($1.99 / $2.49)","Self-serve provisioning (~15 min) for bursts","Owned hardware narrative; no surprise ingress/egress/support fees claimed"],"weaknesses":["Large reserved / Blackwell capacity still sales-gated","Lightning AI merger adds packaging complexity","Less indie DX than RunPod/Modal"],"tags":["gpu","enterprise","capacity"],"scores":{"capabilityFit":8.6,"pricePerformance":7.8,"latencyReliability":8.4,"developerDx":6.8,"privacyPosture":7.8,"lockInRisk":5.5,"maturity":7.8,"overall":7.71,"kind":"curated_estimate","confidence":"high","asOf":"2026-09-14","sources":["Voltage Park H100 pricing: https://voltagepark.com/pricing","Dedicated Reserve Compute: https://www.voltagepark.com/product/dedicated-reserve-compute","Cloud H100 product: https://www.voltagepark.com/product/cloud-h100"],"notes":"Now legitimately self-serve for H100 bursts at published $1.99–$2.49/hr. Prefer RunPod/Nebius for indie experiments; Voltage Park when you want owned-fleet H100 clusters + clear list price without hyperscaler egress theater. Confirm Blackwell quotes separately."},"lastReviewed":"2026-09-14","status":"active"},{"id":"fluidstack","slug":"fluidstack","name":"FluidStack","website":"https://www.fluidstack.io","tagline":"Useful opportunistic H100 capacity if the live console quote clears ~$2.1–2.5/hr. Remain medium until a…","description":"Useful opportunistic H100 capacity if the live console quote clears ~$2.1–2.5/hr. Remain medium until a public SKU table exists. Strengths: Competitive H100/A100/H200 quotes on trackers; Per-minute billing; spot options; Scale narrative for large clusters.","categoryIds":["cat-gpu-cloud"],"jtbdIds":["cheap-gpu","batch-gpu","enterprise-gpu"],"regions":["global","us","eu"],"pricingNotes":"As of 2026-09-14: Official fluidstack.io homepage still has no durable public rate card (capacity/mission positioning). Aggregator trackers disagree: GPUPerHour ~$2.10/hr H100 SXM; gpucloudprices June-2026 ~$2.89; A100 ~$1.30–$1.49; H200 ~$2.30–$3.49. Per-minute billing; spot/reserved exist. Round 4 re-check: still console/AE-quote only. Do not treat tracker midpoints as list price.","strengths":["Competitive H100/A100/H200 quotes on trackers","Per-minute billing; spot options","Scale narrative for large clusters"],"weaknesses":["No durable public rate card on homepage","Marketplace/dedicated mix → variable tenancy","Less indie documentation than RunPod/Vast"],"tags":["gpu","distributed","capacity"],"scores":{"capabilityFit":8,"pricePerformance":8.4,"latencyReliability":7.2,"developerDx":6.8,"privacyPosture":7,"lockInRisk":4.2,"maturity":7.4,"overall":7.46,"kind":"curated_estimate","confidence":"medium","asOf":"2026-09-14","sources":["FluidStack: https://www.fluidstack.io","FluidStack H100 on GPUPerHour: https://gpuperhour.com/providers/fluidstack/h100-sxm-80gb","FluidStack on getdeploying: https://getdeploying.com/fluidstack","GPU Cloud Prices FluidStack: https://gpucloudprices.com/provider/fluidstack/"],"notes":"Useful opportunistic H100 capacity if the live console quote clears ~$2.1–2.5/hr. Remain medium until a public SKU table exists."},"lastReviewed":"2026-09-14","status":"active"},{"id":"paperspace","slug":"paperspace","name":"Paperspace (DigitalOcean)","website":"https://www.paperspace.com","tagline":"Fine for experiments and small fine-tunes. Graduate to RunPod/Lambda/CoreWeave for serious training.","description":"Fine for experiments and small fine-tunes. Graduate to RunPod/Lambda/CoreWeave for serious training. Strengths: Friendly DX for notebooks and small GPUs; DO billing integration; Good for experiments.","categoryIds":["cat-gpu-cloud","cat-cpu-compute"],"jtbdIds":["cheap-gpu","finetune"],"regions":["us","eu"],"pricingNotes":"Gradient notebooks + GPU machines; DigitalOcean-owned; simpler/cheaper entry than hyperscalers.","strengths":["Friendly DX for notebooks and small GPUs","DO billing integration","Good for experiments"],"weaknesses":["Not a frontier multi-node training leader","Hardware freshness lags CoreWeave/Lambda","Production inference better elsewhere"],"tags":["gpu","notebooks","digitalocean"],"scores":{"capabilityFit":7.2,"pricePerformance":7.8,"latencyReliability":7.8,"developerDx":8.2,"privacyPosture":7.2,"lockInRisk":4,"maturity":8.5,"overall":7.59,"kind":"curated_estimate","confidence":"high","asOf":"2026-09-14","sources":["Paperspace: https://www.paperspace.com"],"notes":"Fine for experiments and small fine-tunes. Graduate to RunPod/Lambda/CoreWeave for serious training."},"lastReviewed":"2026-09-14","status":"active"},{"id":"oracle-gpu","slug":"oracle-gpu","name":"Oracle Cloud GPU","website":"https://www.oracle.com/cloud/compute/gpu/","tagline":"Secret weapon for cheap cloud GPUs if you tolerate OCI quirks. Not the agent-native DX winner.","description":"Secret weapon for cheap cloud GPUs if you tolerate OCI quirks. Not the agent-native DX winner. Strengths: Often sharp GPU list prices; Enterprise OCI integration; Good for cost-sensitive cloud shops.","categoryIds":["cat-gpu-cloud"],"jtbdIds":["enterprise-gpu","cheap-gpu","cheap-vms"],"regions":["global"],"pricingNotes":"As of 2026-09-14: OCI GPU instances with historically aggressive promotional pricing vs AWS/Azure/GCP; enterprise contracts and free-tier-adjacent CPU still part of the story. DX and marketplace friction remain the tradeoff vs RunPod/Lambda.","strengths":["Often sharp GPU list prices","Enterprise OCI integration","Good for cost-sensitive cloud shops"],"weaknesses":["DX and AI ecosystem thinner than AWS/GCP","Quota/availability frustration common","Fewer AI-native templates"],"tags":["gpu","oci","enterprise","budget"],"scores":{"capabilityFit":8,"pricePerformance":8.8,"latencyReliability":8,"developerDx":6.5,"privacyPosture":7.8,"lockInRisk":7,"maturity":9,"overall":7.65,"kind":"curated_estimate","confidence":"medium","asOf":"2026-09-14","sources":["Oracle Cloud GPU: https://www.oracle.com/cloud/compute/gpu/","OCI compute pricing: https://www.oracle.com/cloud/price-list/"],"notes":"Secret weapon for cheap cloud GPUs if you tolerate OCI quirks. Not the agent-native DX winner."},"lastReviewed":"2026-09-14","status":"active"},{"id":"aws-gpu","slug":"aws-gpu","name":"AWS GPU (EC2/P5/SageMaker)","website":"https://aws.amazon.com/ec2/instance-types/","tagline":"Use when AWS is destiny. For greenfield agent GPUs, neoclouds win on price and speed-to-pod.","description":"Use when AWS is destiny. For greenfield agent GPUs, neoclouds win on price and speed-to-pod. Strengths: Global regions + enterprise everything; Best-in-class networking options (EFA); Integrates with full AWS estate.","categoryIds":["cat-gpu-cloud","cat-train-finetune"],"jtbdIds":["enterprise-gpu","multi-node-training","production-inference-fleet","batch-gpu"],"regions":["global"],"pricingNotes":"P5/P4d etc. often $4–7+/GPU-hr on-demand; Savings Plans/spot help; SageMaker adds managed tax.","strengths":["Global regions + enterprise everything","Best-in-class networking options (EFA)","Integrates with full AWS estate"],"weaknesses":["Expensive on-demand","Console complexity","Easy to overpay without commits"],"tags":["gpu","aws","enterprise","sagemaker"],"scores":{"capabilityFit":9.4,"pricePerformance":4.8,"latencyReliability":9.5,"developerDx":7,"privacyPosture":9,"lockInRisk":8,"maturity":9.9,"overall":7.72,"kind":"curated_estimate","confidence":"high","asOf":"2026-09-14","sources":["AWS EC2 GPU instances: https://aws.amazon.com/ec2/instance-types/","Amazon SageMaker: https://aws.amazon.com/sagemaker/"],"notes":"Use when AWS is destiny. For greenfield agent GPUs, neoclouds win on price and speed-to-pod."},"lastReviewed":"2026-09-14","status":"active"},{"id":"gcp-gpu","slug":"gcp-gpu","name":"GCP GPU (Compute/GKE)","website":"https://cloud.google.com/compute/docs/gpus","tagline":"Fine inside GCP. Quota tickets will humble you — plan capacity early.","description":"Fine inside GCP. Quota tickets will humble you — plan capacity early. Strengths: Strong GKE + Vertex story; Global infrastructure; Good TPU alternative path for some workloads.","categoryIds":["cat-gpu-cloud"],"jtbdIds":["enterprise-gpu","multi-node-training"],"regions":["global"],"pricingNotes":"A100/H100 via GCE/GKE; committed use discounts; tight Vertex coupling.","strengths":["Strong GKE + Vertex story","Global infrastructure","Good TPU alternative path for some workloads"],"weaknesses":["GPU quota hell is real","On-demand pricing high","Availability uneven by region/SKU"],"tags":["gpu","gcp","gke","enterprise"],"scores":{"capabilityFit":9,"pricePerformance":5,"latencyReliability":9,"developerDx":7.2,"privacyPosture":8.8,"lockInRisk":7.8,"maturity":9.8,"overall":7.59,"kind":"curated_estimate","confidence":"high","asOf":"2026-09-14","sources":["GCP GPUs: https://cloud.google.com/compute/docs/gpus"],"notes":"Fine inside GCP. Quota tickets will humble you — plan capacity early."},"lastReviewed":"2026-09-14","status":"active"},{"id":"azure-gpu","slug":"azure-gpu","name":"Azure GPU (N-series/ND)","website":"https://azure.microsoft.com/products/virtual-machines/","tagline":"Microsoft-shop GPU default. Everyone else should compare RunPod/Lambda first.","description":"Microsoft-shop GPU default. Everyone else should compare RunPod/Lambda first. Strengths: Enterprise Microsoft estate fit; Azure ML integration; Strong compliance portfolio.","categoryIds":["cat-gpu-cloud"],"jtbdIds":["enterprise-gpu","multi-node-training"],"regions":["global"],"pricingNotes":"ND/NC-series GPUs; enterprise agreements; often paired with Azure ML.","strengths":["Enterprise Microsoft estate fit","Azure ML integration","Strong compliance portfolio"],"weaknesses":["Pricing and SKU maze","Quota and regional gaps","Weaker indie AI community vs AWS/GCP"],"tags":["gpu","azure","enterprise"],"scores":{"capabilityFit":8.8,"pricePerformance":5,"latencyReliability":9,"developerDx":6.8,"privacyPosture":8.8,"lockInRisk":7.8,"maturity":9.7,"overall":7.47,"kind":"curated_estimate","confidence":"high","asOf":"2026-09-14","sources":["Azure VMs: https://azure.microsoft.com/products/virtual-machines/"],"notes":"Microsoft-shop GPU default. Everyone else should compare RunPod/Lambda first."},"lastReviewed":"2026-09-14","status":"active"},{"id":"modal","slug":"modal","name":"Modal","website":"https://modal.com","tagline":"Depot darling for Python agent backends and bursty GPU jobs. If you think in functions not VMs, start here.","description":"Depot darling for Python agent backends and bursty GPU jobs. If you think in functions not VMs, start here. Strengths: Best-in-class Python DX for serverless compute; GPUs + sandboxes + cron in one model; Scale-to-zero economics.","categoryIds":["cat-gpu-cloud","cat-cpu-compute","cat-sandboxes-browsers"],"jtbdIds":["serverless-gpu","agent-sandbox","batch-gpu","production-inference-fleet","finetune","agent-backend","code-interpreter","ray-distributed-agents","agent-ephemeral-ci"],"regions":["us","global"],"pricingNotes":"As of 2026-10-24 (modal.com/pricing, fetched 2026-10-24): Python-native serverless GPU/CPU billed per second of compute ('You never pay for idle resources'). GPU list rates per second: H100 SXM5 '$0.001097', H200 SXM '$0.001261', B200 '$0.001736', A100 80 GB '$0.000694', L40S '$0.000542', A10 '$0.000306', L4 '$0.000222', T4 '$0.000164'; CPU '$0.0000131' per core per second (physical core, 2 vCPU equivalent); memory '$0.00000222' per GiB per second. Region selection '1.15 - 1.75x base prices'; non-preemptible execution '3x base prices' ('Sandboxes are billed at non-preemptible rates of 3x list price.'). Containers are kept alive '60 seconds after the last inputs', which is billed (configurable). Plans: Starter '$0 + compute / month' with '$30 / month free credits' ('100 containers + 10 GPU concurrency'); Team '$250 + compute / month' with '$100 / month free credits' ('5000 containers + 50 GPU concurrency'); Enterprise custom. Shared Endpoints are 'billed per token rather than per second' and not covered by included compute ('This took effect September 1, 2026.'). Excellent DX for jobs and endpoints; cold starts matter for interactive agents.","strengths":["Best-in-class Python DX for serverless compute","GPUs + sandboxes + cron in one model","Scale-to-zero economics"],"weaknesses":["Premium vs raw RunPod $/GPU-sec sometimes","Python-centric (polyglot harder)","Less classic VM control"],"tags":["serverless","gpu","python","sandbox","batch"],"scores":{"capabilityFit":9,"pricePerformance":8.2,"latencyReliability":8.6,"developerDx":9.6,"privacyPosture":7.8,"lockInRisk":4.5,"maturity":8.2,"overall":8.44,"kind":"curated_estimate","confidence":"high","asOf":"2026-10-19","sources":["Modal: https://modal.com","Modal pricing: https://modal.com/pricing"],"notes":"Depot darling for Python agent backends and bursty GPU jobs. If you think in functions not VMs, start here."},"lastReviewed":"2026-10-19","status":"active"},{"id":"replicate","slug":"replicate","name":"Replicate","website":"https://replicate.com","tagline":"Ship a model demo by lunch. Migrate hot paths to Modal/RunPod/Fireworks when the bill hurts.","description":"Ship a model demo by lunch. Migrate hot paths to Modal/RunPod/Fireworks when the bill hurts. Strengths: Easiest model zoo deployment; Great for demos and media models; Huge public model catalog.","categoryIds":["cat-gpu-cloud","cat-media-gen","cat-inference-apis"],"jtbdIds":["serverless-gpu","media-generation","open-model-inference","image-generation","3d-generation"],"regions":["global"],"pricingNotes":"Run Cog models per-second; prediction-based billing; higher effective GPU cost, lowest friction.","strengths":["Easiest model zoo deployment","Great for demos and media models","Huge public model catalog"],"weaknesses":["Expensive at scale vs self-serve GPUs","Less control over infra","Cold starts on unpopular models"],"tags":["serverless","cog","media","models"],"scores":{"capabilityFit":8.4,"pricePerformance":6,"latencyReliability":8.2,"developerDx":9.4,"privacyPosture":7,"lockInRisk":4,"maturity":8.8,"overall":7.82,"kind":"curated_estimate","confidence":"high","asOf":"2026-09-14","sources":["Replicate: https://replicate.com"],"notes":"Ship a model demo by lunch. Migrate hot paths to Modal/RunPod/Fireworks when the bill hurts."},"lastReviewed":"2026-09-14","status":"active"},{"id":"beam","slug":"beam","name":"Beam","website":"https://www.beam.cloud","tagline":"Viable Modal/RunPod alternative for serverless endpoints. Evaluate cold starts on your workload.","description":"Viable Modal/RunPod alternative for serverless endpoints. Evaluate cold starts on your workload. Strengths: Simple serverless ML deploys; Competitive with early Banana-like UX; Containers + functions.","categoryIds":["cat-gpu-cloud","cat-cpu-compute"],"jtbdIds":["serverless-gpu"],"regions":["us","global"],"pricingNotes":"As of 2026-09-14 (beam.cloud/pricing live): Developer **$0/mo + usage**; Team **$89/mo + usage** (3 seats, +$25/seat); Growth contact. Page headline: sandboxes from **$0.319/hr**, H100s from **$1.83/hr**. Example CPU calc: 2 cores (4 vCPU) $0.09 + 8 GiB $0.06 = $0.15/hr. GPU concurrency 5/50/1000+ by tier. BYOC management fee $0.019/vCPU-hr + $0.009/GB-RAM-hr (compute on your cloud bill). SOC2/HIPAA on higher tiers.","strengths":["Simple serverless ML deploys","Competitive with early Banana-like UX","Containers + functions"],"weaknesses":["Smaller ecosystem than Modal/Replicate","Mindshare battle ongoing","Fewer public benchmarks"],"tags":["serverless","gpu","containers"],"scores":{"capabilityFit":7.8,"pricePerformance":8,"latencyReliability":7.5,"developerDx":8.2,"privacyPosture":7,"lockInRisk":4,"maturity":7,"overall":7.57,"kind":"curated_estimate","confidence":"high","asOf":"2026-09-14","sources":["Beam pricing: https://www.beam.cloud/pricing","Beam: https://www.beam.cloud","Beam docs: https://docs.beam.cloud"],"notes":"Viable Modal/RunPod alternative for serverless endpoints. Evaluate cold starts on your workload."},"lastReviewed":"2026-09-14","status":"active"},{"id":"banana","slug":"banana","name":"Banana.dev (sunset)","website":"https://www.banana.dev","tagline":"Dead. If a tutorial says Banana, translate to RunPod Serverless or Modal. Included so Depot can warn agents…","description":"Dead. If a tutorial says Banana, translate to RunPod Serverless or Modal. Included so Depot can warn agents away. Strengths: Pioneered accessible serverless GPUs; Simple app.py deploy mental model.","categoryIds":["cat-gpu-cloud"],"jtbdIds":[],"regions":["us"],"pricingNotes":"HISTORICAL: Serverless GPU platform shut down March 31, 2024. Do not build new systems on Banana.","strengths":["Pioneered accessible serverless GPUs","Simple app.py deploy mental model"],"weaknesses":["Discontinued — no production use","Migration required to RunPod/Modal/etc.","Brand still appears in stale tutorials"],"tags":["gpu","serverless","sunset","historical"],"scores":{"capabilityFit":1,"pricePerformance":2,"latencyReliability":0.5,"developerDx":4,"privacyPosture":4,"lockInRisk":9,"maturity":1.5,"overall":1.91,"kind":"curated_estimate","confidence":"high","asOf":"2026-09-14","sources":["Sunsetting Serverless GPUs - Banana: https://www.banana.dev/blog/sunset","RunPod Banana migration: https://www.runpod.io/blog/banana-dev-migration-runpod"],"notes":"Dead. If a tutorial says Banana, translate to RunPod Serverless or Modal. Included so Depot can warn agents away."},"lastReviewed":"2026-09-14","status":"deprecated"},{"id":"fly-io","slug":"fly-io","name":"Fly.io","website":"https://fly.io","tagline":"Excellent for globally distributed agent sidecars and Machines-based sandboxes. Prefer Railway/Render if you…","description":"Excellent for globally distributed agent sidecars and Machines-based sandboxes. Prefer Railway/Render if you want less infra thinking. Strengths: Global edge VMs close to users; Machines API great for ephemeral agent workloads; WireGuard private networking.","categoryIds":["cat-cpu-compute","cat-sandboxes-browsers"],"jtbdIds":["edge-deploy","agent-sandbox","app-hosting","agent-backend"],"regions":["global"],"pricingNotes":"Pay for machines/volumes/bandwidth; Firecracker VMs; scale-to-zero; Machines API for agents.","strengths":["Global edge VMs close to users","Machines API great for ephemeral agent workloads","WireGuard private networking"],"weaknesses":["Ops complexity vs PaaS","Support historically uneven","Pricing surprises on egress/volumes"],"tags":["paas","edge","machines","firecracker"],"scores":{"capabilityFit":8.8,"pricePerformance":8.2,"latencyReliability":8,"developerDx":8.8,"privacyPosture":7.8,"lockInRisk":3.5,"maturity":8.5,"overall":8.28,"kind":"curated_estimate","confidence":"high","asOf":"2026-09-14","sources":["Fly.io: https://fly.io","Fly Machines: https://fly.io/docs/machines/"],"notes":"Excellent for globally distributed agent sidecars and Machines-based sandboxes. Prefer Railway/Render if you want less infra thinking."},"lastReviewed":"2026-09-14","status":"active"},{"id":"railway","slug":"railway","name":"Railway","website":"https://railway.app","tagline":"Depot favorite for shipping agent APIs quickly. When bills climb, move steady-state to Fly/Hetzner/DO.","description":"Depot favorite for shipping agent APIs quickly. When bills climb, move steady-state to Fly/Hetzner/DO. Strengths: Superb DX for full-stack agent backends; Databases + services colocated; Fast iteration.","categoryIds":["cat-cpu-compute"],"jtbdIds":["app-hosting","agent-backend","persistent-dev-env","app-backend"],"regions":["global","us","eu"],"pricingNotes":"Usage-based PaaS; services/databases; simple git deploys; hobby→pro plans.","strengths":["Superb DX for full-stack agent backends","Databases + services colocated","Fast iteration"],"weaknesses":["Less edge footprint than Fly/CF","Can get pricey at scale vs raw VMs","Enterprise features thinner than Render/AWS"],"tags":["paas","dx","postgres","deploy"],"scores":{"capabilityFit":8.2,"pricePerformance":7.8,"latencyReliability":8.4,"developerDx":9.4,"privacyPosture":7.2,"lockInRisk":4,"maturity":8.2,"overall":8.08,"kind":"curated_estimate","confidence":"high","asOf":"2026-09-14","sources":["Railway: https://railway.app"],"notes":"Depot favorite for shipping agent APIs quickly. When bills climb, move steady-state to Fly/Hetzner/DO."},"lastReviewed":"2026-09-14","status":"active"},{"id":"render","slug":"render","name":"Render","website":"https://render.com","tagline":"Boring reliable PaaS. Great when Railway feels too hip and AWS feels too heavy.","description":"Boring reliable PaaS. Great when Railway feels too hip and AWS feels too heavy. Strengths: Polished PaaS with predictable mental model; Good private networking and preview envs; Solid Postgres story.","categoryIds":["cat-cpu-compute"],"jtbdIds":["app-hosting","agent-backend"],"regions":["global","us","eu"],"pricingNotes":"Web services, background workers, Postgres, cron; fixed + usage elements; free tier limited.","strengths":["Polished PaaS with predictable mental model","Good private networking and preview envs","Solid Postgres story"],"weaknesses":["Less cutting-edge than Fly Machines for agents","Cold starts on free/low tiers","GPU not the point"],"tags":["paas","postgres","workers"],"scores":{"capabilityFit":8,"pricePerformance":7.5,"latencyReliability":8.8,"developerDx":9,"privacyPosture":7.8,"lockInRisk":4,"maturity":8.8,"overall":8.1,"kind":"curated_estimate","confidence":"high","asOf":"2026-09-14","sources":["Render: https://render.com"],"notes":"Boring reliable PaaS. Great when Railway feels too hip and AWS feels too heavy."},"lastReviewed":"2026-09-14","status":"active"},{"id":"digitalocean","slug":"digitalocean","name":"DigitalOcean","website":"https://www.digitalocean.com","tagline":"Classic simple cloud. Use for steady VMs and Spaces; pair Paperspace/RunPod for GPUs.","description":"Classic simple cloud. Use for steady VMs and Spaces; pair Paperspace/RunPod for GPUs. Strengths: Clear pricing and docs; Good enough for most agent backends; Spaces S3-compat.","categoryIds":["cat-cpu-compute","cat-object-storage","cat-gpu-cloud"],"jtbdIds":["app-hosting","cheap-vms","object-storage","cheap-egress-storage"],"regions":["global","us","eu","ap"],"pricingNotes":"Droplets, App Platform, Spaces, Managed DBs, Paperspace GPUs; simple pricing.","strengths":["Clear pricing and docs","Good enough for most agent backends","Spaces S3-compat"],"weaknesses":["Not the cheapest EU option (Hetzner)","AI-native features limited","Support tiered"],"tags":["iaas","paas","spaces","droplets"],"scores":{"capabilityFit":7.8,"pricePerformance":8.2,"latencyReliability":8.8,"developerDx":8.5,"privacyPosture":7.5,"lockInRisk":3.5,"maturity":9.2,"overall":8.15,"kind":"curated_estimate","confidence":"high","asOf":"2026-09-14","sources":["DigitalOcean: https://www.digitalocean.com"],"notes":"Classic simple cloud. Use for steady VMs and Spaces; pair Paperspace/RunPod for GPUs."},"lastReviewed":"2026-09-14","status":"active"},{"id":"hetzner","slug":"hetzner","name":"Hetzner","website":"https://www.hetzner.com","tagline":"When you need cheap always-on CPU in EU, start here. DIY the agent orchestration yourself.","description":"When you need cheap always-on CPU in EU, start here. DIY the agent orchestration yourself. Strengths: Outstanding price/performance for CPU; EU data center strength; Transparent pricing.","categoryIds":["cat-cpu-compute","cat-object-storage"],"jtbdIds":["cheap-vms","object-storage","cheap-egress-storage"],"regions":["eu","us"],"pricingNotes":"Very cheap cloud VMs and dedicated; auction servers; Object Storage; EU-strong.","strengths":["Outstanding price/performance for CPU","EU data center strength","Transparent pricing"],"weaknesses":["Fewer regions than hyperscalers","Less managed PaaS candy","Support model is ticket-oriented"],"tags":["iaas","budget","eu","bare-metal"],"scores":{"capabilityFit":7.5,"pricePerformance":9.6,"latencyReliability":8.6,"developerDx":7.2,"privacyPosture":8,"lockInRisk":2.5,"maturity":9,"overall":8.2,"kind":"curated_estimate","confidence":"high","asOf":"2026-09-14","sources":["Hetzner Cloud: https://www.hetzner.com/cloud","Hetzner: https://www.hetzner.com"],"notes":"When you need cheap always-on CPU in EU, start here. DIY the agent orchestration yourself."},"lastReviewed":"2026-09-14","status":"active"},{"id":"cloudflare-workers","slug":"cloudflare-workers","name":"Cloudflare Workers","website":"https://workers.cloudflare.com","tagline":"Best edge runtime for lightweight agent gateways and tools. Heavy LLM loops still belong on Modal/Fly/GPU…","description":"Best edge runtime for lightweight agent gateways and tools. Heavy LLM loops still belong on Modal/Fly/GPU hosts. Strengths: Insane global edge; R2 zero-egress narrative; Durable Objects for agent state.","categoryIds":["cat-cpu-compute"],"jtbdIds":["edge-deploy","agent-backend","agent-memory-kv","mcp-server-hosting"],"regions":["global"],"pricingNotes":"Request-based Workers + Durable Objects + R2 + Queues; generous free tier; CPU time limits.","strengths":["Insane global edge","R2 zero-egress narrative","Durable Objects for agent state"],"weaknesses":["CPU/time limits constrain heavy agent loops","Vendor-specific programming model","Debugging distributed edge is hard"],"tags":["edge","serverless","r2","durable-objects"],"scores":{"capabilityFit":8.6,"pricePerformance":9.2,"latencyReliability":9.2,"developerDx":8.5,"privacyPosture":8,"lockInRisk":5.5,"maturity":9.5,"overall":8.53,"kind":"curated_estimate","confidence":"high","asOf":"2026-09-14","sources":["Cloudflare Workers: https://workers.cloudflare.com","Cloudflare AI: https://developers.cloudflare.com/workers-ai/"],"notes":"Best edge runtime for lightweight agent gateways and tools. Heavy LLM loops still belong on Modal/Fly/GPU hosts."},"lastReviewed":"2026-09-14","status":"active"},{"id":"vercel","slug":"vercel","name":"Vercel","website":"https://vercel.com","tagline":"Ship the agent UI and thin API routes here. Put long-running agent workers on Modal/Fly/Railway.","description":"Ship the agent UI and thin API routes here. Put long-running agent workers on Modal/Fly/Railway. Strengths: Best Next.js / AI SDK hosting DX; Preview deployments; Edge + serverless functions.","categoryIds":["cat-cpu-compute"],"jtbdIds":["app-hosting","edge-deploy"],"regions":["global"],"pricingNotes":"Frontend/serverless PaaS; Fluid compute; AI SDK integration; usage-based; great Next.js DX.","strengths":["Best Next.js / AI SDK hosting DX","Preview deployments","Edge + serverless functions"],"weaknesses":["Serverless timeouts/cost for long agents","Can get expensive","Not for GPU work"],"tags":["paas","nextjs","ai-sdk","serverless"],"scores":{"capabilityFit":8.4,"pricePerformance":7,"latencyReliability":9,"developerDx":9.6,"privacyPosture":7.5,"lockInRisk":6,"maturity":9.2,"overall":8.1,"kind":"curated_estimate","confidence":"high","asOf":"2026-09-14","sources":["Vercel: https://vercel.com","Vercel AI SDK: https://sdk.vercel.ai"],"notes":"Ship the agent UI and thin API routes here. Put long-running agent workers on Modal/Fly/Railway."},"lastReviewed":"2026-09-14","status":"active"},{"id":"aws-ec2","slug":"aws-ec2","name":"AWS EC2","website":"https://aws.amazon.com/ec2/","tagline":"Default enterprise VM substrate. Agents should wrap it — humans shouldn't live in the console.","description":"Default enterprise VM substrate. Agents should wrap it — humans shouldn't live in the console. Strengths: Infinite instance types and regions; Enterprise controls; Spot for cheap batch.","categoryIds":["cat-cpu-compute"],"jtbdIds":["cheap-vms","enterprise-compute","agent-backend","postgres-vector"],"regions":["global"],"pricingNotes":"Broadest instance catalog; Spot/Savings Plans; everything else in AWS orbits EC2.","strengths":["Infinite instance types and regions","Enterprise controls","Spot for cheap batch"],"weaknesses":["Complexity tax","Easy to misconfigure security/cost","DX raw vs PaaS"],"tags":["iaas","aws","vms","spot"],"scores":{"capabilityFit":9.2,"pricePerformance":7,"latencyReliability":9.5,"developerDx":6.5,"privacyPosture":8.8,"lockInRisk":7.5,"maturity":9.9,"overall":8.01,"kind":"curated_estimate","confidence":"high","asOf":"2026-09-14","sources":["Amazon EC2: https://aws.amazon.com/ec2/"],"notes":"Default enterprise VM substrate. Agents should wrap it — humans shouldn't live in the console."},"lastReviewed":"2026-09-14","status":"active"},{"id":"gcp-compute","slug":"gcp-compute","name":"GCP Compute Engine","website":"https://cloud.google.com/products/compute","tagline":"Solid when you're already on GCP. Otherwise Hetzner/DO/Fly are faster paths for agents.","description":"Solid when you're already on GCP. Otherwise Hetzner/DO/Fly are faster paths for agents. Strengths: Flexible machine shapes; Strong networking/data stack adjacency; Preemptible savings.","categoryIds":["cat-cpu-compute"],"jtbdIds":["enterprise-compute","cheap-vms"],"regions":["global"],"pricingNotes":"Custom machine types; sustained use discounts; preemptible/Spot VMs.","strengths":["Flexible machine shapes","Strong networking/data stack adjacency","Preemptible savings"],"weaknesses":["Quota and IAM complexity","Fewer 'just deploy' PaaS vibes","Support via Cloud Customer Care tiers"],"tags":["iaas","gcp","vms"],"scores":{"capabilityFit":8.8,"pricePerformance":7.2,"latencyReliability":9.2,"developerDx":6.8,"privacyPosture":8.8,"lockInRisk":7.5,"maturity":9.8,"overall":7.93,"kind":"curated_estimate","confidence":"high","asOf":"2026-09-14","sources":["Compute Engine: https://cloud.google.com/products/compute"],"notes":"Solid when you're already on GCP. Otherwise Hetzner/DO/Fly are faster paths for agents."},"lastReviewed":"2026-09-14","status":"active"},{"id":"azure-vms","slug":"azure-vms","name":"Azure Virtual Machines","website":"https://azure.microsoft.com/products/virtual-machines/","tagline":"Enterprise Microsoft default. Indie agent stacks should look elsewhere first.","description":"Enterprise Microsoft default. Indie agent stacks should look elsewhere first. Strengths: Microsoft enterprise integration; Hybrid cloud story; Compliance portfolio.","categoryIds":["cat-cpu-compute"],"jtbdIds":["enterprise-compute"],"regions":["global"],"pricingNotes":"Broad VM families; hybrid AD integration; enterprise agreements.","strengths":["Microsoft enterprise integration","Hybrid cloud story","Compliance portfolio"],"weaknesses":["Portal complexity","Pricing opacity","Weaker indie agent community"],"tags":["iaas","azure","vms"],"scores":{"capabilityFit":8.6,"pricePerformance":6.5,"latencyReliability":9.2,"developerDx":6.2,"privacyPosture":8.8,"lockInRisk":7.8,"maturity":9.8,"overall":7.64,"kind":"curated_estimate","confidence":"high","asOf":"2026-09-14","sources":["Azure VMs: https://azure.microsoft.com/products/virtual-machines/"],"notes":"Enterprise Microsoft default. Indie agent stacks should look elsewhere first."},"lastReviewed":"2026-09-14","status":"active"},{"id":"pinecone","slug":"pinecone","name":"Pinecone","website":"https://www.pinecone.io","tagline":"Still the zero-ops vector default. If you already run Postgres, try pgvector before paying Pinecone rents.","description":"Still the zero-ops vector default. If you already run Postgres, try pgvector before paying Pinecone rents. Strengths: Zero-ops managed vector DB; Mature product and ecosystem; Serverless scales with RAG apps.","categoryIds":["cat-vector-memory"],"jtbdIds":["managed-vector","rag-memory","rag-enterprise","billion-scale-vector","prototype-rag"],"regions":["us","eu"],"pricingNotes":"Serverless + dedicated/pod tiers; usage on storage/reads/writes; Dedicated Read Nodes for high QPS.","strengths":["Zero-ops managed vector DB","Mature product and ecosystem","Serverless scales with RAG apps"],"weaknesses":["Costly at large scale vs pgvector/Qdrant","Proprietary — export takes work","Overkill under ~millions of vectors for many teams"],"tags":["vector","managed","serverless","rag"],"scores":{"capabilityFit":9,"pricePerformance":6.2,"latencyReliability":9,"developerDx":8.8,"privacyPosture":8,"lockInRisk":7,"maturity":9,"overall":7.95,"kind":"curated_estimate","confidence":"high","asOf":"2026-09-14","sources":["Pinecone: https://www.pinecone.io"],"notes":"Still the zero-ops vector default. If you already run Postgres, try pgvector before paying Pinecone rents."},"lastReviewed":"2026-09-14","status":"active"},{"id":"weaviate","slug":"weaviate","name":"Weaviate","website":"https://weaviate.io","tagline":"Best when hybrid search and modularity matter. Cloud for convenience; self-host when lock-in anxiety spikes.","description":"Best when hybrid search and modularity matter. Cloud for convenience; self-host when lock-in anxiety spikes. Strengths: Hybrid search + modules; Open-source escape hatch; Strong RAG feature set.","categoryIds":["cat-vector-memory"],"jtbdIds":["managed-vector","hybrid-search-rag","self-host-vector","rag-enterprise"],"regions":["global","us","eu"],"pricingNotes":"Open-source + Weaviate Cloud; modular vectorizers; hybrid search; usage/cluster pricing on cloud.","strengths":["Hybrid search + modules","Open-source escape hatch","Strong RAG feature set"],"weaknesses":["Ops complexity self-hosted","Cloud pricing needs care","Performance tuning required at scale"],"tags":["vector","hybrid-search","oss","rag"],"scores":{"capabilityFit":8.8,"pricePerformance":7.5,"latencyReliability":8.4,"developerDx":8.2,"privacyPosture":8.2,"lockInRisk":3.5,"maturity":8.8,"overall":8.2,"kind":"curated_estimate","confidence":"high","asOf":"2026-09-14","sources":["Weaviate: https://weaviate.io"],"notes":"Best when hybrid search and modularity matter. Cloud for convenience; self-host when lock-in anxiety spikes."},"lastReviewed":"2026-09-14","status":"active"},{"id":"qdrant","slug":"qdrant","name":"Qdrant","website":"https://qdrant.tech","tagline":"Depot pick for serious vector workloads with filters. Self-host or Cloud — either way you can leave.","description":"Depot pick for serious vector workloads with filters. Self-host or Cloud — either way you can leave. Strengths: Top filtered-search performance; Great self-host story; Cloud option without giving up OSS.","categoryIds":["cat-vector-memory"],"jtbdIds":["self-host-vector","managed-vector","rag-memory","cheap-vector-scale","billion-scale-vector","hybrid-search-rag"],"regions":["global","eu","us"],"pricingNotes":"Rust engine; open-source + Qdrant Cloud; excellent filtered ANN; competitive cloud pricing.","strengths":["Top filtered-search performance","Great self-host story","Cloud option without giving up OSS"],"weaknesses":["Ecosystem slightly smaller than Pinecone mindshare","Need to tune HNSW for best results","UI/ops polish varying by tier"],"tags":["vector","rust","oss","filtering","rag"],"scores":{"capabilityFit":9.2,"pricePerformance":8.8,"latencyReliability":8.6,"developerDx":8.4,"privacyPosture":8.2,"lockInRisk":3,"maturity":8.5,"overall":8.59,"kind":"curated_estimate","confidence":"high","asOf":"2026-10-19","sources":["Qdrant: https://qdrant.tech"],"notes":"Depot pick for serious vector workloads with filters. Self-host or Cloud — either way you can leave."},"lastReviewed":"2026-10-19","status":"active"},{"id":"chroma","slug":"chroma","name":"Chroma","website":"https://www.trychroma.com","tagline":"Prototype on Chroma; graduate to Qdrant/pgvector/Pinecone before your corpus and QPS grow up.","description":"Prototype on Chroma; graduate to Qdrant/pgvector/Pinecone before your corpus and QPS grow up. Strengths: Fastest path to local RAG prototype; Python-native DX; Good for demos and notebooks.","categoryIds":["cat-vector-memory"],"jtbdIds":["prototype-rag","rag-memory","self-host-vector","embedded-vector"],"regions":["global"],"pricingNotes":"As of 2026-09-14: Open-source embeddable vector DB + hosted Chroma Cloud; free/local for prototypes; Cloud usage-based for managed. Excellent DX for demos; graduate before high QPS/filtered production.","strengths":["Fastest path to local RAG prototype","Python-native DX","Good for demos and notebooks"],"weaknesses":["Not the scale leader vs Qdrant/Milvus/Pinecone","Production features still catching up","Hosted maturity evolving"],"tags":["vector","oss","prototype","python"],"scores":{"capabilityFit":7.2,"pricePerformance":9,"latencyReliability":7,"developerDx":9.2,"privacyPosture":7.5,"lockInRisk":2.5,"maturity":7.5,"overall":7.88,"kind":"curated_estimate","confidence":"high","asOf":"2026-09-14","sources":["Chroma: https://www.trychroma.com","Chroma docs: https://docs.trychroma.com","Chroma Cloud: https://www.trychroma.com/cloud"],"notes":"Prototype on Chroma; graduate to Qdrant/pgvector/Pinecone before your corpus and QPS grow up."},"lastReviewed":"2026-09-14","status":"active"},{"id":"turbopuffer","slug":"turbopuffer","name":"Turbopuffer","website":"https://turbopuffer.com","tagline":"When your vector bill scares you and much of the corpus is cold, Turbopuffer is the interesting answer.","description":"When your vector bill scares you and much of the corpus is cold, Turbopuffer is the interesting answer. Strengths: Excellent cost for large-scale / cold-heavy retrieval; Modern architecture for agent memory at scale; Simple operational model.","categoryIds":["cat-vector-memory","cat-object-storage"],"jtbdIds":["cheap-vector-scale","rag-memory","billion-scale-vector","embedded-vector","vector-object-hybrid-memory"],"regions":["us","global"],"pricingNotes":"As of 2026-10-24 (turbopuffer.com/pricing, fetched 2026-10-24): Usage-based with plan minimums: launch '$16/month minimum usage', scale '$256/month minimum usage', enterprise '>= $4,096/month' with '35% usage premium'. 'You'll pay for actual usage beyond the minimum.' 'Unused amounts don't roll over to the next month.' Calculator rates: 'Storage ≤ $0.33 per GB', 'Writes ≤ $2.00 per GB', 'Queries ≤ $1.00 per PB', billed on logical bytes. Queries: 'an 80% marginal discount on data queried between 32 GB and 128 GB, and a 96% marginal discount on data queried beyond 128 GB'; writes: 'Every write is billed at least 10 KB', batch write discount up to 50% ('reaches 50% at approximately 3.2 MB'). Scale adds 'HIPAA-ready BAA' and 'Single Sign-On (SSO)'; enterprise adds 'Uptime SLA 99.95%' and 'BYOC'. No free tier ('we have chosen a commercial-only model').","strengths":["Excellent cost for large-scale / cold-heavy retrieval","Modern architecture for agent memory at scale","Simple operational model"],"weaknesses":["Younger than Pinecone/Weaviate","Feature set narrower for fancy hybrid modules","Less community content"],"tags":["vector","object-storage","scale","cost"],"scores":{"capabilityFit":8.6,"pricePerformance":9.4,"latencyReliability":8.2,"developerDx":8.5,"privacyPosture":7.8,"lockInRisk":4.5,"maturity":7,"overall":8.21,"kind":"curated_estimate","confidence":"high","asOf":"2026-09-14","sources":["turbopuffer pricing: https://turbopuffer.com/pricing","turbopuffer pricing changelog: https://turbopuffer.com/docs/pricing-log","Turbopuffer: https://turbopuffer.com","Turbopuffer docs: https://turbopuffer.com/docs"],"notes":"When your vector bill scares you and much of the corpus is cold, Turbopuffer is the interesting answer."},"lastReviewed":"2026-09-14","status":"active"},{"id":"supabase-pgvector","slug":"supabase-pgvector","name":"Supabase (pgvector)","website":"https://supabase.com","tagline":"Depot default for most agent apps under ~25–50M vectors. Don't add Pinecone until Postgres actually hurts.","description":"Depot default for most agent apps under ~25–50M vectors. Don't add Pinecone until Postgres actually hurts. Strengths: One backend for auth, DB, storage, vectors; Best DX for many Next.js agent apps; Open standards (Postgres).","categoryIds":["cat-vector-memory","cat-auth-payments","cat-object-storage"],"jtbdIds":["postgres-vector","rag-memory","app-backend","cheap-vector-scale","prototype-rag","agent-memory-kv","auth","user-management","object-storage","serverless-db","hybrid-search-rag"],"regions":["global"],"pricingNotes":"Postgres + pgvector + Auth + Storage; usage on compute/storage; free tier. Vectors hitchhike on Postgres.","strengths":["One backend for auth, DB, storage, vectors","Best DX for many Next.js agent apps","Open standards (Postgres)"],"weaknesses":["Vector scale ceiling vs dedicated engines","Need pgvectorscale/careful indexing past tens of M","Not a pure vector specialist"],"tags":["postgres","pgvector","baas","rag","auth"],"scores":{"capabilityFit":8.5,"pricePerformance":9,"latencyReliability":8.6,"developerDx":9.4,"privacyPosture":8,"lockInRisk":3,"maturity":9,"overall":8.64,"kind":"curated_estimate","confidence":"high","asOf":"2026-09-14","sources":["Supabase: https://supabase.com","pgvector: https://github.com/pgvector/pgvector"],"notes":"Depot default for most agent apps under ~25–50M vectors. Don't add Pinecone until Postgres actually hurts."},"lastReviewed":"2026-09-14","status":"active"},{"id":"redis","slug":"redis","name":"Redis","website":"https://redis.io","tagline":"Use as agent hot memory/cache, not as your only 100M-vector archive.","description":"Use as agent hot memory/cache, not as your only 100M-vector archive. Strengths: Ultra-low latency KV + vectors; Ubiquitous in stacks already; Great session/memory cache for agents.","categoryIds":["cat-vector-memory"],"jtbdIds":["agent-memory-kv","rag-memory"],"regions":["global"],"pricingNotes":"Redis OSS + Redis Cloud; vector search via Redis Query Engine; pay for memory-centric hosting.","strengths":["Ultra-low latency KV + vectors","Ubiquitous in stacks already","Great session/memory cache for agents"],"weaknesses":["RAM cost at large vector corpora","Not ideal as sole long-term vector warehouse","Cluster ops complexity"],"tags":["kv","cache","vector","memory"],"scores":{"capabilityFit":8.2,"pricePerformance":7,"latencyReliability":9,"developerDx":8.5,"privacyPosture":7.5,"lockInRisk":4,"maturity":9.6,"overall":8.07,"kind":"curated_estimate","confidence":"high","asOf":"2026-09-14","sources":["Redis: https://redis.io","Redis Cloud: https://redis.io/cloud/"],"notes":"Use as agent hot memory/cache, not as your only 100M-vector archive."},"lastReviewed":"2026-09-14","status":"active"},{"id":"mongodb-atlas-vector","slug":"mongodb-atlas-vector","name":"MongoDB Atlas Vector Search","website":"https://www.mongodb.com/products/platform/atlas-vector-search","tagline":"Pick if Mongo is already your system of record. Otherwise pgvector/Qdrant are cleaner agent defaults.","description":"Pick if Mongo is already your system of record. Otherwise pgvector/Qdrant are cleaner agent defaults. Strengths: Vectors next to operational documents; Managed Atlas reliability; Familiar for Mongo shops.","categoryIds":["cat-vector-memory"],"jtbdIds":["managed-vector","rag-memory"],"regions":["global"],"pricingNotes":"Vector search inside Atlas; billed with Atlas cluster; good if data already in Mongo.","strengths":["Vectors next to operational documents","Managed Atlas reliability","Familiar for Mongo shops"],"weaknesses":["Not the ANN performance leader in isolation","Cost tied to Atlas sizing","Overkill if you only need vectors"],"tags":["mongodb","vector","atlas","rag"],"scores":{"capabilityFit":8,"pricePerformance":7.2,"latencyReliability":8.8,"developerDx":8,"privacyPosture":8,"lockInRisk":5.5,"maturity":9.4,"overall":7.87,"kind":"curated_estimate","confidence":"high","asOf":"2026-09-14","sources":["Atlas Vector Search: https://www.mongodb.com/products/platform/atlas-vector-search"],"notes":"Pick if Mongo is already your system of record. Otherwise pgvector/Qdrant are cleaner agent defaults."},"lastReviewed":"2026-09-14","status":"active"},{"id":"zilliz","slug":"zilliz","name":"Zilliz (Milvus)","website":"https://zilliz.com","tagline":"When you're truly at massive scale, Milvus/Zilliz earns its complexity. Don't start here for a chatbot MVP.","description":"When you're truly at massive scale, Milvus/Zilliz earns its complexity. Don't start here for a chatbot MVP. Strengths: Billion-scale vector DNA; Open-source Milvus core; Serious indexing options.","categoryIds":["cat-vector-memory"],"jtbdIds":["billion-scale-vector","self-host-vector","managed-vector","cheap-vector-scale","hybrid-search-rag"],"regions":["global","us","cn"],"pricingNotes":"Managed Milvus (Zilliz Cloud); open-source Milvus for self-host; designed for large-scale ANN.","strengths":["Billion-scale vector DNA","Open-source Milvus core","Serious indexing options"],"weaknesses":["Heavier ops than Pinecone/Qdrant Cloud","Complexity for small RAG apps","Learning curve"],"tags":["vector","milvus","scale","oss"],"scores":{"capabilityFit":9.2,"pricePerformance":7.8,"latencyReliability":8.4,"developerDx":7.2,"privacyPosture":7.8,"lockInRisk":3.5,"maturity":8.8,"overall":8.16,"kind":"curated_estimate","confidence":"high","asOf":"2026-09-14","sources":["Zilliz: https://zilliz.com","Milvus: https://milvus.io"],"notes":"When you're truly at massive scale, Milvus/Zilliz earns its complexity. Don't start here for a chatbot MVP."},"lastReviewed":"2026-09-14","status":"active"},{"id":"langsmith","slug":"langsmith","name":"LangSmith","website":"https://www.langchain.com/langsmith","tagline":"If you live in LangGraph, LangSmith is the obvious observatory. Otherwise compare Langfuse/Helicone first.","description":"If you live in LangGraph, LangSmith is the obvious observatory. Otherwise compare Langfuse/Helicone first. Strengths: Best tracing UX in LangChain ecosystem; Datasets/evals integrated; Production debugging for chains/agents.","categoryIds":["cat-observability"],"jtbdIds":["llm-tracing","evals","agent-trace-session-replay"],"regions":["us","global"],"pricingNotes":"Seat + trace volume pricing; deep LangChain/LangGraph integration; Plus tiers from tens of $/mo.","strengths":["Best tracing UX in LangChain ecosystem","Datasets/evals integrated","Production debugging for chains/agents"],"weaknesses":["LangChain-centric gravity (works elsewhere but shines in-family)","Cost scales with trace volume","Lock-in to LangSmith workflows"],"tags":["tracing","evals","langchain","agents"],"scores":{"capabilityFit":9,"pricePerformance":7,"latencyReliability":8.8,"developerDx":9,"privacyPosture":7.8,"lockInRisk":6.5,"maturity":8.8,"overall":8.09,"kind":"curated_estimate","confidence":"high","asOf":"2026-09-14","sources":["LangSmith: https://www.langchain.com/langsmith"],"notes":"If you live in LangGraph, LangSmith is the obvious observatory. Otherwise compare Langfuse/Helicone first."},"lastReviewed":"2026-09-14","status":"active"},{"id":"helicone","slug":"helicone","name":"Helicone","website":"https://www.helicone.ai","tagline":"Maintenance mode since 2026-03-03 (Mintlify); site now says 'no longer part of Mintlify'. Runner-up only, not a pick.","description":"In maintenance mode since 2026-03-03 per Mintlify's acquisition post (https://www.mintlify.com/blog/mintlify-acquires-helicone); helicone.ai now shows 'Helicone is no longer part of Mintlify. Details to come soon.' (re-checked 2026-10-21). Replaced as the cost-monitoring and llm-rate-limit-shield pick on 2026-10-21 (by Langfuse and Cloudflare AI Gateway); kept as a runner-up with the maintenance_mode flag. Former description: Fastest observability win for OpenAI-compat stacks. Put Helicone in front, ask questions later. Strengths: Drop-in proxy for OpenAI-compat providers; Cost and latency analytics; Fast to adopt.","categoryIds":["cat-observability"],"jtbdIds":["llm-tracing","cost-monitoring","agent-cloud-finops"],"regions":["global"],"pricingNotes":"Proxy/gateway logging; usage-based plans from low tens of $/mo; open-source components.","strengths":["Drop-in proxy for OpenAI-compat providers","Cost and latency analytics","Fast to adopt"],"weaknesses":["Deeper agent-graph evals behind Braintrust/LangSmith","Proxy hop considerations","Feature race with peers"],"tags":["gateway","tracing","cost","proxy","maintenance-mode"],"scores":{"capabilityFit":8.2,"pricePerformance":8.5,"latencyReliability":8.4,"developerDx":8.8,"privacyPosture":7.5,"lockInRisk":4,"maturity":8,"overall":8.13,"kind":"curated_estimate","confidence":"high","asOf":"2026-10-19","sources":["Helicone: https://www.helicone.ai"],"notes":"Fastest observability win for OpenAI-compat stacks. Put Helicone in front, ask questions later."},"lastReviewed":"2026-10-19","status":"active"},{"id":"braintrust","slug":"braintrust","name":"Braintrust","website":"https://www.braintrust.dev","tagline":"Depot pick when evals are the product quality gate. Pair with Helicone/Langfuse for cheap request logs if…","description":"Depot pick when evals are the product quality gate. Pair with Helicone/Langfuse for cheap request logs if needed. Strengths: Best-in-class eval workflows; Great for prompt/agent regression gates; Generous seat model.","categoryIds":["cat-observability"],"jtbdIds":["evals","llm-tracing","oss-evals","production-guardrail-evals","continuous-eval-ci"],"regions":["us","global"],"pricingNotes":"Eval-first platform; data-based pricing with unlimited seats on many plans; strong CI eval story.","strengths":["Best-in-class eval workflows","Great for prompt/agent regression gates","Generous seat model"],"weaknesses":["Broader tracing competitors exist","Price entry higher than hobby tools","Learning curve for eval design"],"tags":["evals","tracing","ci","prompts"],"scores":{"capabilityFit":9.2,"pricePerformance":7.5,"latencyReliability":8.6,"developerDx":8.8,"privacyPosture":8,"lockInRisk":5,"maturity":8.2,"overall":8.23,"kind":"curated_estimate","confidence":"high","asOf":"2026-10-19","sources":["Braintrust: https://www.braintrust.dev"],"notes":"Depot pick when evals are the product quality gate. Pair with Helicone/Langfuse for cheap request logs if needed."},"lastReviewed":"2026-10-19","status":"active"},{"id":"arize-phoenix","slug":"arize-phoenix","name":"Arize Phoenix","website":"https://phoenix.arize.com","tagline":"Choose Phoenix when you want open-source observability you can own. Cloud Arize when enterprise wants a…","description":"Choose Phoenix when you want open-source observability you can own. Cloud Arize when enterprise wants a vendor. Strengths: OSS-first tracing/evals; Strong embeddings/drift heritage from Arize; Avoid SaaS lock-in via self-host.","categoryIds":["cat-observability"],"jtbdIds":["llm-tracing","evals","experiment-tracking","infra-observability"],"regions":["global"],"pricingNotes":"Open-source Phoenix free to self-host; Arize AX cloud paid; OpenTelemetry-friendly.","strengths":["OSS-first tracing/evals","Strong embeddings/drift heritage from Arize","Avoid SaaS lock-in via self-host"],"weaknesses":["Self-host ops burden","UX polish varies vs LangSmith","Cloud packaging can confuse"],"tags":["oss","tracing","evals","otel"],"scores":{"capabilityFit":8.6,"pricePerformance":9,"latencyReliability":8.2,"developerDx":8,"privacyPosture":8.5,"lockInRisk":2.5,"maturity":8.6,"overall":8.44,"kind":"curated_estimate","confidence":"high","asOf":"2026-09-14","sources":["Arize Phoenix: https://phoenix.arize.com","Arize: https://arize.com"],"notes":"Choose Phoenix when you want open-source observability you can own. Cloud Arize when enterprise wants a vendor."},"lastReviewed":"2026-09-14","status":"active"},{"id":"langfuse","slug":"langfuse","name":"Langfuse","website":"https://langfuse.com","tagline":"Depot default observability for most agent teams: open, affordable, good enough to serious.","description":"Depot default observability for most agent teams: open, affordable, good enough to serious. Strengths: OSS + cloud flexibility; Excellent agent/trace UI for the price; EU posture.","categoryIds":["cat-observability"],"jtbdIds":["llm-tracing","evals","cost-monitoring","experiment-tracking","infra-observability"],"regions":["eu","global"],"pricingNotes":"Open-source + cloud; generous self-host; cloud from ~$29–59/mo tiers; EU-friendly.","strengths":["OSS + cloud flexibility","Excellent agent/trace UI for the price","EU posture"],"weaknesses":["Enterprise features chase LangSmith","You operate it if self-hosting","Ecosystem still growing"],"tags":["oss","tracing","evals","eu"],"scores":{"capabilityFit":8.8,"pricePerformance":9.2,"latencyReliability":8.4,"developerDx":8.8,"privacyPosture":8.8,"lockInRisk":2.5,"maturity":8.2,"overall":8.66,"kind":"curated_estimate","confidence":"high","asOf":"2026-09-14","sources":["Langfuse: https://langfuse.com"],"notes":"Depot default observability for most agent teams: open, affordable, good enough to serious."},"lastReviewed":"2026-09-14","status":"active"},{"id":"weights-biases","slug":"weights-biases","name":"Weights & Biases","website":"https://wandb.ai","tagline":"Use for training/fine-tune discipline. For pure agent request tracing, Langfuse/Helicone are leaner.","description":"Use for training/fine-tune discipline. For pure agent request tracing, Langfuse/Helicone are leaner. Strengths: Gold standard ML experiment tracking; Expanding LLM/weave tooling; Enterprise adoption.","categoryIds":["cat-observability","cat-train-finetune"],"jtbdIds":["experiment-tracking","evals","ml-dataset-versioning"],"regions":["global"],"pricingNotes":"Experiment tracking + W&B Weave/LLM tools; seat and usage based; free tiers for individuals.","strengths":["Gold standard ML experiment tracking","Expanding LLM/weave tooling","Enterprise adoption"],"weaknesses":["Heavier than needed for simple API traces","Cost at team scale","Historically training-centric"],"tags":["mlops","experiments","weave","training"],"scores":{"capabilityFit":8.8,"pricePerformance":6.8,"latencyReliability":9,"developerDx":8.2,"privacyPosture":7.8,"lockInRisk":5.5,"maturity":9.4,"overall":8.04,"kind":"curated_estimate","confidence":"high","asOf":"2026-09-14","sources":["Weights & Biases: https://wandb.ai"],"notes":"Use for training/fine-tune discipline. For pure agent request tracing, Langfuse/Helicone are leaner."},"lastReviewed":"2026-09-14","status":"active"},{"id":"honeycomb","slug":"honeycomb","name":"Honeycomb","website":"https://www.honeycomb.io","tagline":"When your agent outage is an infra mystery, Honeycomb shines. Still pair with an LLM-native tracer for…","description":"When your agent outage is an infra mystery, Honeycomb shines. Still pair with an LLM-native tracer for prompts. Strengths: Best-in-class wide-events debugging; Superb for distributed agent infra; Strong engineering culture/docs.","categoryIds":["cat-observability"],"jtbdIds":["infra-observability","llm-tracing"],"regions":["us","global"],"pricingNotes":"Event-based observability pricing; excellent high-cardinality querying; not LLM-specialized.","strengths":["Best-in-class wide-events debugging","Superb for distributed agent infra","Strong engineering culture/docs"],"weaknesses":["Not LLM-eval native","Need instrumentation work","Price for high event volume"],"tags":["otel","tracing","events","sre"],"scores":{"capabilityFit":8.5,"pricePerformance":7,"latencyReliability":9.2,"developerDx":8.5,"privacyPosture":8,"lockInRisk":4.5,"maturity":9,"overall":8.13,"kind":"curated_estimate","confidence":"high","asOf":"2026-09-14","sources":["Honeycomb: https://www.honeycomb.io"],"notes":"When your agent outage is an infra mystery, Honeycomb shines. Still pair with an LLM-native tracer for prompts."},"lastReviewed":"2026-09-14","status":"active"},{"id":"browserbase","slug":"browserbase","name":"Browserbase","website":"https://www.browserbase.com","tagline":"Depot default for agents that need a real browser. Pair with Stagehand/Browser Use for the reasoning loop.","description":"Depot default for agents that need a real browser. Pair with Stagehand/Browser Use for the reasoning loop. Strengths: Cleanest agent-oriented browser cloud; Playwright/Puppeteer compatible; Session persistence + stealth features.","categoryIds":["cat-sandboxes-browsers","cat-search-scrape"],"jtbdIds":["cloud-browser","web-agent","web-scrape","enterprise-proxy","computer-use-browser","browser-agent-loop"],"regions":["us","global"],"pricingNotes":"Session-based cloud browsers; free/$20/$99/custom tiers; Playwright-compatible; recordings/stealth.","strengths":["Cleanest agent-oriented browser cloud","Playwright/Puppeteer compatible","Session persistence + stealth features"],"weaknesses":["Cost at high parallel sessions","Still need your own agent logic","Less self-host than Steel"],"tags":["browser","playwright","agents","stealth"],"scores":{"capabilityFit":9.2,"pricePerformance":7.5,"latencyReliability":8.8,"developerDx":9.4,"privacyPosture":7.5,"lockInRisk":5,"maturity":8.2,"overall":8.3,"kind":"curated_estimate","confidence":"high","asOf":"2026-10-19","sources":["Browserbase: https://www.browserbase.com","Browserbase pricing: https://www.browserbase.com/pricing"],"notes":"Depot default for agents that need a real browser. Pair with Stagehand/Browser Use for the reasoning loop."},"lastReviewed":"2026-10-19","status":"active"},{"id":"steel-dev","slug":"steel-dev","name":"Steel.dev","website":"https://steel.dev","tagline":"Pick Steel when you need to self-host browsers or hate SaaS lock-in. Browserbase wins on polish.","description":"Pick Steel when you need to self-host browsers or hate SaaS lock-in. Browserbase wins on polish. Strengths: Self-hostable browser infra; Agent-focused API; Escape hatch from pure SaaS.","categoryIds":["cat-sandboxes-browsers","cat-search-scrape"],"jtbdIds":["cloud-browser","self-host-browser","web-agent"],"regions":["global"],"pricingNotes":"As of 2026-09-14 (docs.steel.dev/overview/pricinglimits): Launch **$0 + usage** ($30 one-time free credits/90d); Scale **$250 + usage** ($100 credits/mo); Enterprise custom. Browser hours Launch **$0.10/hr** / Scale **$0.08/hr** (per-minute round-up). Proxy $10/$6 per GB; CAPTCHA $3/$1 per 1k; Browser tools (/scrape,/screenshot,/pdf) $5/1k. Concurrent sessions 10 / 100 / 1000+. Max session 15m / 1h / 24h. OSS self-host escape hatch (steel-browser). Legacy Starter/Developer/Pro still for existing customers.","strengths":["Self-hostable browser infra","Agent-focused API","Escape hatch from pure SaaS"],"weaknesses":["Ecosystem smaller than Browserbase","You own ops if self-hosting","Stealth arms race ongoing"],"tags":["browser","oss","self-host","agents"],"scores":{"capabilityFit":8.6,"pricePerformance":8.5,"latencyReliability":8,"developerDx":8.6,"privacyPosture":8.5,"lockInRisk":2.5,"maturity":7.2,"overall":8.27,"kind":"curated_estimate","confidence":"high","asOf":"2026-09-14","sources":["Steel pricing/limits: https://docs.steel.dev/overview/pricinglimits","Steel.dev: https://steel.dev","Steel docs: https://docs.steel.dev","Steel GitHub: https://github.com/steel-dev/steel-browser"],"notes":"Pick Steel when you need to self-host browsers or hate SaaS lock-in. Browserbase wins on polish."},"lastReviewed":"2026-09-14","status":"active"},{"id":"browserless","slug":"browserless","name":"Browserless","website":"https://www.browserless.io","tagline":"Veteran choice for headless fleets. Great if you already speak Playwright fleets, not just agents.","description":"Veteran choice for headless fleets. Great if you already speak Playwright fleets, not just agents. Strengths: Mature browser-as-a-service; BrowserQL for concise scraping; Self-host path.","categoryIds":["cat-sandboxes-browsers","cat-search-scrape"],"jtbdIds":["cloud-browser","self-host-browser","web-agent"],"regions":["global","us","eu"],"pricingNotes":"Hosted Playwright/Puppeteer/BrowserQL; usage and concurrent connection plans; self-host options.","strengths":["Mature browser-as-a-service","BrowserQL for concise scraping","Self-host path"],"weaknesses":["Less 'agent product' packaging than Browserbase","Pricing on concurrency can spike","You still write automation"],"tags":["browser","playwright","browserql","scraping"],"scores":{"capabilityFit":8.8,"pricePerformance":7.8,"latencyReliability":8.6,"developerDx":8.5,"privacyPosture":7.5,"lockInRisk":3.5,"maturity":8.8,"overall":8.25,"kind":"curated_estimate","confidence":"high","asOf":"2026-09-14","sources":["Browserless: https://www.browserless.io"],"notes":"Veteran choice for headless fleets. Great if you already speak Playwright fleets, not just agents."},"lastReviewed":"2026-09-14","status":"active"},{"id":"scrapingbee","slug":"scrapingbee","name":"ScrapingBee","website":"https://www.scrapingbee.com","tagline":"Use for fetch-the-HTML jobs. For multi-step login/click agents, use Browserbase/Steel.","description":"Use for fetch-the-HTML jobs. For multi-step login/click agents, use Browserbase/Steel. Strengths: Simple scrape API; Handles proxies/JS rendering; Fast integration.","categoryIds":["cat-search-scrape"],"jtbdIds":["web-scrape","serp-scrape","llm-ready-extract","enterprise-proxy"],"regions":["global"],"pricingNotes":"Credit-based HTML/API scraping with JS rendering and proxy; plans by credit packs.","strengths":["Simple scrape API","Handles proxies/JS rendering","Fast integration"],"weaknesses":["Not a full interactive browser agent platform","Credit math can confuse","Less control than Browserbase sessions"],"tags":["scrape","proxy","html","serp"],"scores":{"capabilityFit":7.8,"pricePerformance":8.2,"latencyReliability":8.4,"developerDx":8.8,"privacyPosture":7,"lockInRisk":4,"maturity":8.5,"overall":7.98,"kind":"curated_estimate","confidence":"high","asOf":"2026-09-14","sources":["ScrapingBee: https://www.scrapingbee.com"],"notes":"Use for fetch-the-HTML jobs. For multi-step login/click agents, use Browserbase/Steel."},"lastReviewed":"2026-09-14","status":"active"},{"id":"firecrawl","slug":"firecrawl","name":"Firecrawl","website":"https://www.firecrawl.dev","tagline":"Depot pick for turning the web into LLM context. Browserbase when you need to *act*, Firecrawl when you need…","description":"Depot pick for turning the web into LLM context. Browserbase when you need to *act*, Firecrawl when you need to *read*. Strengths: LLM-ready output (markdown/structured); Excellent agent/MCP integrations; Search + extract in one vendor.","categoryIds":["cat-search-scrape"],"jtbdIds":["web-scrape","llm-ready-extract","search-augmented-llm","web-agent"],"regions":["global"],"pricingNotes":"Crawl/scrape to clean markdown/JSON for LLMs; API credits; agent/sandbox features expanding.","strengths":["LLM-ready output (markdown/structured)","Excellent agent/MCP integrations","Search + extract in one vendor"],"weaknesses":["Cost vs raw scrape APIs at huge volume","Less interactive than full browser clouds","Newer than classic scrape vendors"],"tags":["scrape","crawl","markdown","llm","mcp"],"scores":{"capabilityFit":9,"pricePerformance":8,"latencyReliability":8.4,"developerDx":9.4,"privacyPosture":7.2,"lockInRisk":4.5,"maturity":7.8,"overall":8.25,"kind":"curated_estimate","confidence":"high","asOf":"2026-10-19","sources":["Firecrawl: https://www.firecrawl.dev"],"notes":"Depot pick for turning the web into LLM context. Browserbase when you need to *act*, Firecrawl when you need to *read*."},"lastReviewed":"2026-10-19","status":"active"},{"id":"bright-data","slug":"bright-data","name":"Bright Data","website":"https://brightdata.com","tagline":"Nuclear option for hard anti-bot targets. Start with Firecrawl/ScrapingBee; escalate here when blocked.","description":"Nuclear option for hard anti-bot targets. Start with Firecrawl/ScrapingBee; escalate here when blocked. Strengths: Largest proxy/residential network; Enterprise compliance options; SERP and unlocker products.","categoryIds":["cat-search-scrape"],"jtbdIds":["web-scrape","serp-scrape","enterprise-proxy"],"regions":["global"],"pricingNotes":"Enterprise proxy/scraping network; pay for bandwidth/success; Web Scraper IDE; SERP APIs.","strengths":["Largest proxy/residential network","Enterprise compliance options","SERP and unlocker products"],"weaknesses":["Expensive and complex for indie agents","Ethics/ToS considerations on residential","Overkill for simple pages"],"tags":["proxy","serp","enterprise","scrape"],"scores":{"capabilityFit":9.2,"pricePerformance":5.5,"latencyReliability":9,"developerDx":7,"privacyPosture":7,"lockInRisk":5,"maturity":9.2,"overall":7.66,"kind":"curated_estimate","confidence":"high","asOf":"2026-09-14","sources":["Bright Data: https://brightdata.com"],"notes":"Nuclear option for hard anti-bot targets. Start with Firecrawl/ScrapingBee; escalate here when blocked."},"lastReviewed":"2026-09-14","status":"active"},{"id":"playwright","slug":"playwright","name":"Playwright (library)","website":"https://playwright.dev","tagline":"The API you probably already use — under Browserbase/Steel/Browserless. Listed so agents don't confuse…","description":"The API you probably already use — under Browserbase/Steel/Browserless. Listed so agents don't confuse library with cloud. Strengths: Industry-standard automation API; Multi-browser; Zero license cost.","categoryIds":["cat-sandboxes-browsers"],"jtbdIds":["self-host-browser"],"regions":["n/a"],"pricingNotes":"Free open-source browser automation library (Microsoft). Hosting/compute not included.","strengths":["Industry-standard automation API","Multi-browser","Zero license cost"],"weaknesses":["Not a hosted provider — you run it","Anti-bot and scaling are your problem","Ops burden for fleets"],"tags":["library","browser","oss","microsoft"],"scores":{"capabilityFit":9.5,"pricePerformance":9.5,"latencyReliability":7,"developerDx":9.2,"privacyPosture":9.5,"lockInRisk":1,"maturity":9.5,"overall":9.05,"kind":"curated_estimate","confidence":"high","asOf":"2026-09-14","sources":["Playwright: https://playwright.dev"],"notes":"The API you probably already use — under Browserbase/Steel/Browserless. Listed so agents don't confuse library with cloud."},"lastReviewed":"2026-09-14","status":"active"},{"id":"puppeteer","slug":"puppeteer","name":"Puppeteer (library)","website":"https://pptr.dev","tagline":"Still fine for Chrome-only scripts. New agent projects should prefer Playwright unless stuck on Puppeteer.","description":"Still fine for Chrome-only scripts. New agent projects should prefer Playwright unless stuck on Puppeteer. Strengths: Mature Chrome automation; Huge ecosystem; Free.","categoryIds":["cat-sandboxes-browsers"],"jtbdIds":["self-host-browser"],"regions":["n/a"],"pricingNotes":"Free OSS Chrome DevTools protocol automation (Google lineage). Hosting not included.","strengths":["Mature Chrome automation","Huge ecosystem","Free"],"weaknesses":["Chrome-centric vs Playwright multi-browser","Not hosted","Fleet ops DIY"],"tags":["library","browser","oss","chrome"],"scores":{"capabilityFit":8.8,"pricePerformance":9.5,"latencyReliability":6.8,"developerDx":8.8,"privacyPosture":9.5,"lockInRisk":1,"maturity":9.6,"overall":8.79,"kind":"curated_estimate","confidence":"high","asOf":"2026-09-14","sources":["Puppeteer: https://pptr.dev"],"notes":"Still fine for Chrome-only scripts. New agent projects should prefer Playwright unless stuck on Puppeteer."},"lastReviewed":"2026-09-14","status":"active"},{"id":"e2b","slug":"e2b","name":"E2B","website":"https://e2b.dev","tagline":"Depot default code sandbox for agents. Use for untrusted code; Daytona when you need persistent…","description":"Depot default code sandbox for agents. Use for untrusted code; Daytona when you need persistent devworkspaces. Also the depot default Linux desktop sandbox for computer-use (e2b-desktop + VNC stream) when you already standardize on E2B. Strengths: Purpose-built for LLM code execution; Strong isolation (microVMs); Excellent agent SDK fit.","categoryIds":["cat-sandboxes-browsers"],"jtbdIds":["agent-sandbox","code-interpreter","persistent-dev-env","desktop-computer-use"],"regions":["us","eu","global"],"pricingNotes":"Firecracker microVM sandboxes billed per second; templates; code interpreter for agents.","strengths":["Purpose-built for LLM code execution","Strong isolation (microVMs)","Excellent agent SDK fit"],"weaknesses":["Per-sandbox-second costs add up","Less persistent than Daytona workspaces","Younger than hyperscalers"],"tags":["sandbox","firecracker","code-interpreter","agents","desktop","computer-use","vnc"],"scores":{"capabilityFit":9.4,"pricePerformance":8,"latencyReliability":8.6,"developerDx":9.4,"privacyPosture":8.5,"lockInRisk":4,"maturity":8,"overall":8.56,"kind":"curated_estimate","confidence":"high","asOf":"2026-09-14","sources":["E2B: https://e2b.dev","E2B computer use docs: https://www.e2b.dev/docs/use-cases/computer-use"],"notes":"Depot default code sandbox for agents. Use for untrusted code; Daytona when you need persistent devworkspaces. Also the depot default Linux desktop sandbox for computer-use (e2b-desktop + VNC stream) when you already standardize on E2B."},"lastReviewed":"2026-09-14","status":"active"},{"id":"daytona","slug":"daytona","name":"Daytona","website":"https://daytona.io","tagline":"Choose when agents need sticky filesystems and full dev environments. E2B for disposable code exec. Confirm…","description":"Choose when agents need sticky filesystems and full dev environments. E2B for disposable code exec. Confirm licensing after 2026 closed-source shift. Strengths: Persistent coding environments for agents; OSS + cloud; Fast workspace spin-up.","categoryIds":["cat-sandboxes-browsers","cat-cpu-compute"],"jtbdIds":["agent-sandbox","persistent-dev-env","platform-sandbox"],"regions":["global"],"pricingNotes":"As of 2026-09-14 (daytona.io/pricing live): PAYG per-second — vCPU **$0.0504/h**, Memory **$0.0162/GiB-h**, Storage **$0.000108/GiB-h** (first 5 GiB free). $200 free compute signup. Windows $0.0858/vCPU-h. GPU preemptible $/hour (stored rows, reconfirmed 2026-10-27): H100 $2.27, H200 $2.61, B200 $3.59, RTX 4090 $0.57, RTX 5090 $0.74, RTX PRO 6000 $1.74. GPU on-demand $/hour (stored rows since 2026-10-27): H100 $3.95, H200 $4.54, B200 $6.25, B300 $6.25, MI355X $5.99, RTX 4090 $0.99, RTX 5090 $1.29, RTX PRO 6000 $3.03. Startup program up to $50k credits. Closed-source cloud mid-2026 — verify OSS assumptions.","strengths":["Persistent coding environments for agents","OSS + cloud","Fast workspace spin-up"],"weaknesses":["Container isolation ≠ Firecracker","Positioning overlaps E2B/Modal","Ecosystem still scaling"],"tags":["sandbox","workspaces","oss","agents"],"scores":{"capabilityFit":8.8,"pricePerformance":8.2,"latencyReliability":8.2,"developerDx":8.8,"privacyPosture":8,"lockInRisk":3,"maturity":7.2,"overall":8.24,"kind":"curated_estimate","confidence":"high","asOf":"2026-09-14","sources":["Daytona pricing: https://www.daytona.io/pricing","Daytona billing docs: https://www.daytona.io/docs/en/billing/","Daytona: https://daytona.io","Daytona docs: https://www.daytona.io/docs","Agent sandbox landscape 2026: https://shipengtao.com/en/agent-sandbox/"],"notes":"Choose when agents need sticky filesystems and full dev environments. E2B for disposable code exec. Confirm licensing after 2026 closed-source shift."},"lastReviewed":"2026-09-14","status":"active"},{"id":"fly-machines","slug":"fly-machines","name":"Fly Machines","website":"https://fly.io/docs/machines/","tagline":"DIY sandbox substrate with global edge. Prefer E2B unless you need custom machine control.","description":"DIY sandbox substrate with global edge. Prefer E2B unless you need custom machine control. Strengths: Programmable VMs globally; Firecracker isolation; Integrate with Fly networking.","categoryIds":["cat-sandboxes-browsers","cat-cpu-compute"],"jtbdIds":["agent-sandbox","edge-deploy","persistent-dev-env"],"regions":["global"],"pricingNotes":"API-driven Firecracker VMs on Fly; pay while running; great for custom sandboxes.","strengths":["Programmable VMs globally","Firecracker isolation","Integrate with Fly networking"],"weaknesses":["You build more sandbox UX yourself","Not agent-specialized like E2B","Ops knowledge required"],"tags":["sandbox","vms","fly","firecracker"],"scores":{"capabilityFit":8.5,"pricePerformance":8.5,"latencyReliability":8.2,"developerDx":8,"privacyPosture":7.8,"lockInRisk":3,"maturity":8.5,"overall":8.21,"kind":"curated_estimate","confidence":"high","asOf":"2026-09-14","sources":["Fly Machines: https://fly.io/docs/machines/"],"notes":"DIY sandbox substrate with global edge. Prefer E2B unless you need custom machine control."},"lastReviewed":"2026-09-14","status":"active"},{"id":"deno-deploy","slug":"deno-deploy","name":"Deno Deploy","website":"https://deno.com/deploy","tagline":"Nice for TS edge tools. Don't confuse isolate deploy with E2B-class untrusted code sandboxes.","description":"Nice for TS edge tools. Don't confuse isolate deploy with E2B-class untrusted code sandboxes. Strengths: Superb TS/edge DX; Fast global deploys; Secure-by-default isolates.","categoryIds":["cat-cpu-compute","cat-sandboxes-browsers"],"jtbdIds":["edge-deploy","agent-backend"],"regions":["global"],"pricingNotes":"V8 isolate edge deploy; usage-based; TypeScript-native; limited for heavy sandboxing.","strengths":["Superb TS/edge DX","Fast global deploys","Secure-by-default isolates"],"weaknesses":["Not a full VM sandbox for untrusted agent code","CPU limits","Niche vs CF Workers mindshare"],"tags":["edge","typescript","isolates","deno"],"scores":{"capabilityFit":7.8,"pricePerformance":8.5,"latencyReliability":8.6,"developerDx":9,"privacyPosture":8.2,"lockInRisk":4.5,"maturity":8,"overall":8.13,"kind":"curated_estimate","confidence":"high","asOf":"2026-09-14","sources":["Deno Deploy: https://deno.com/deploy"],"notes":"Nice for TS edge tools. Don't confuse isolate deploy with E2B-class untrusted code sandboxes."},"lastReviewed":"2026-09-14","status":"active"},{"id":"resend","slug":"resend","name":"Resend","website":"https://resend.com","tagline":"Depot default for app/agent transactional email. Migrate to Postmark/SendGrid only for specific…","description":"Depot default for app/agent transactional email. Migrate to Postmark/SendGrid only for specific deliverability needs. Strengths: Best DX for modern app email; React Email; Simple API agents love.","categoryIds":["cat-comms"],"jtbdIds":["transactional-email","marketing-email","notification-orchestration"],"regions":["global"],"pricingNotes":"Modern transactional email API; generous free tier; usage-based; React Email ecosystem.","strengths":["Best DX for modern app email","React Email","Simple API agents love"],"weaknesses":["Deliverability still your domain reputation","Fewer legacy enterprise features than SendGrid","SMS not the focus"],"tags":["email","transactional","react-email"],"scores":{"capabilityFit":8.8,"pricePerformance":8.8,"latencyReliability":8.6,"developerDx":9.6,"privacyPosture":8,"lockInRisk":3.5,"maturity":8,"overall":8.57,"kind":"curated_estimate","confidence":"high","asOf":"2026-09-14","sources":["Resend: https://resend.com"],"notes":"Depot default for app/agent transactional email. Migrate to Postmark/SendGrid only for specific deliverability needs."},"lastReviewed":"2026-09-14","status":"active"},{"id":"sendgrid","slug":"sendgrid","name":"SendGrid (Twilio)","website":"https://sendgrid.com","tagline":"Enterprise volume email incumbent. New agent apps should try Resend/Postmark first.","description":"Enterprise volume email incumbent. New agent apps should try Resend/Postmark first. Strengths: Scale and ecosystem maturity; Marketing + transactional; Twilio adjacency.","categoryIds":["cat-comms"],"jtbdIds":["transactional-email","marketing-email"],"regions":["global"],"pricingNotes":"High-volume email; free tier limited; ESSP/pro plans; Twilio-owned.","strengths":["Scale and ecosystem maturity","Marketing + transactional","Twilio adjacency"],"weaknesses":["DX feels dated vs Resend","Support/pricing complaints common","Complex for simple agent mail"],"tags":["email","twilio","marketing","transactional"],"scores":{"capabilityFit":9,"pricePerformance":7.5,"latencyReliability":8.8,"developerDx":7.2,"privacyPosture":7.8,"lockInRisk":5,"maturity":9.5,"overall":8.08,"kind":"curated_estimate","confidence":"high","asOf":"2026-09-14","sources":["SendGrid: https://sendgrid.com"],"notes":"Enterprise volume email incumbent. New agent apps should try Resend/Postmark first."},"lastReviewed":"2026-09-14","status":"active"},{"id":"postmark","slug":"postmark","name":"Postmark","website":"https://postmarkapp.com","tagline":"When email *must* land (receipts, auth), Postmark is the grown-up pick. Resend for DX; Postmark for…","description":"When email *must* land (receipts, auth), Postmark is the grown-up pick. Resend for DX; Postmark for deliverability paranoia. Strengths: Deliverability-first culture; Clear transactional focus; Great activity UI/API.","categoryIds":["cat-comms"],"jtbdIds":["transactional-email"],"regions":["global"],"pricingNotes":"Transactional-focused email; message-based pricing; excellent deliverability reputation.","strengths":["Deliverability-first culture","Clear transactional focus","Great activity UI/API"],"weaknesses":["Not for marketing blasts","Price vs Resend for tiny apps","Smaller hipster mindshare"],"tags":["email","transactional","deliverability"],"scores":{"capabilityFit":9,"pricePerformance":8.2,"latencyReliability":9.2,"developerDx":8.8,"privacyPosture":8.2,"lockInRisk":3.5,"maturity":9,"overall":8.6,"kind":"curated_estimate","confidence":"high","asOf":"2026-09-14","sources":["Postmark: https://postmarkapp.com"],"notes":"When email *must* land (receipts, auth), Postmark is the grown-up pick. Resend for DX; Postmark for deliverability paranoia."},"lastReviewed":"2026-09-14","status":"active"},{"id":"twilio","slug":"twilio","name":"Twilio","website":"https://www.twilio.com","tagline":"Default when agents need SMS/voice globally. For email-only, don't drag Twilio in.","description":"Default when agents need SMS/voice globally. For email-only, don't drag Twilio in. Strengths: Ubiquitous programmable comms; Global numbers coverage; Verify + Conversations + voice.","categoryIds":["cat-comms"],"jtbdIds":["sms","voice","whatsapp","notification-orchestration","voice-agents","agent-pstn-telephony"],"regions":["global"],"pricingNotes":"Pay-per-message/minute across SMS/voice/WhatsApp; number rental; Verify add-ons.","strengths":["Ubiquitous programmable comms","Global numbers coverage","Verify + Conversations + voice"],"weaknesses":["Can get expensive","Complexity of products","Competitors cheaper for SMS-only"],"tags":["sms","voice","whatsapp","verify"],"scores":{"capabilityFit":9.4,"pricePerformance":6.5,"latencyReliability":9.2,"developerDx":8.5,"privacyPosture":8,"lockInRisk":5.5,"maturity":9.7,"overall":8.26,"kind":"curated_estimate","confidence":"high","asOf":"2026-10-19","sources":["Twilio: https://www.twilio.com"],"notes":"Default when agents need SMS/voice globally. For email-only, don't drag Twilio in."},"lastReviewed":"2026-10-19","status":"active"},{"id":"courier","slug":"courier","name":"Courier","website":"https://www.courier.com","tagline":"Use when your agent/product notifies across many channels with preferences. Skip if you only send email.","description":"Use when your agent/product notifies across many channels with preferences. Skip if you only send email. Strengths: One API for multi-channel notify; Template/preference center; Great for product notification matrices.","categoryIds":["cat-comms"],"jtbdIds":["notification-orchestration","marketing-email","sms","whatsapp"],"regions":["global"],"pricingNotes":"Notification orchestration across email/SMS/push/Slack; usage + seat elements.","strengths":["One API for multi-channel notify","Template/preference center","Great for product notification matrices"],"weaknesses":["Another layer on top of Resend/Twilio","Cost adds up","Overkill for single-channel"],"tags":["notifications","orchestration","multi-channel"],"scores":{"capabilityFit":8.8,"pricePerformance":7.2,"latencyReliability":8.6,"developerDx":9,"privacyPosture":7.8,"lockInRisk":5,"maturity":8.2,"overall":8.09,"kind":"curated_estimate","confidence":"high","asOf":"2026-09-14","sources":["Courier: https://www.courier.com"],"notes":"Use when your agent/product notifies across many channels with preferences. Skip if you only send email."},"lastReviewed":"2026-09-14","status":"active"},{"id":"clerk","slug":"clerk","name":"Clerk","website":"https://clerk.com","tagline":"Depot default auth for agent products with users. WorkOS when enterprise SSO is the main job.","description":"Depot default auth for agent products with users. WorkOS when enterprise SSO is the main job. Strengths: Best-in-class auth DX for modern apps; Prebuilt components agents/docs love; Organizations/B2B features.","categoryIds":["cat-auth-payments"],"jtbdIds":["auth","user-management","enterprise-sso","app-backend"],"regions":["global"],"pricingNotes":"Bills monthly retained users (MRU: a user who returns at least one day after signing up), not MAU (https://clerk.com/pricing, fetched 2026-10-21): Hobby free up to 50,000 MRU per app; Pro $25/mo and Business $300/mo include 50,000 MRU per app, then graduated $0.02 / $0.018 / $0.015 / $0.012 per MRU; 1 SSO connection included on paid plans, then $75 / $60 each (101+ contact us). B2B Authentication add-on $100/mo; beautiful prebuilt UIs.","strengths":["Best-in-class auth DX for modern apps","Prebuilt components agents/docs love","Organizations/B2B features"],"weaknesses":["MAU pricing anxiety at scale","Less 'enterprise IdP' than Auth0/WorkOS for some cases","Migration off Clerk takes work"],"tags":["auth","mau","b2b","nextjs"],"scores":{"capabilityFit":9,"pricePerformance":8,"latencyReliability":9,"developerDx":9.8,"privacyPosture":8.2,"lockInRisk":6.5,"maturity":8.5,"overall":8.43,"kind":"curated_estimate","confidence":"high","asOf":"2026-10-19","sources":["Clerk: https://clerk.com"],"notes":"Depot default auth for agent products with users. WorkOS when enterprise SSO is the main job."},"lastReviewed":"2026-10-19","status":"active"},{"id":"auth0","slug":"auth0","name":"Auth0 (Okta)","website":"https://auth0.com","tagline":"Enterprise identity Swiss Army knife. For startups, Clerk/WorkOS usually hurt less.","description":"Enterprise identity Swiss Army knife. For startups, Clerk/WorkOS usually hurt less. Strengths: Enterprise identity depth; Huge ecosystem/extensibility; Compliance pedigree.","categoryIds":["cat-auth-payments"],"jtbdIds":["auth","enterprise-sso","user-management"],"regions":["global"],"pricingNotes":"MAU and enterprise contracts; extensive IdP ecosystem; Okta-owned.","strengths":["Enterprise identity depth","Huge ecosystem/extensibility","Compliance pedigree"],"weaknesses":["Pricing and complexity notorious","DX heavier than Clerk","Easy to overbuy"],"tags":["auth","okta","enterprise","oidc"],"scores":{"capabilityFit":9.2,"pricePerformance":5.5,"latencyReliability":9.2,"developerDx":7.5,"privacyPosture":8.8,"lockInRisk":7,"maturity":9.6,"overall":7.85,"kind":"curated_estimate","confidence":"high","asOf":"2026-09-14","sources":["Auth0: https://auth0.com"],"notes":"Enterprise identity Swiss Army knife. For startups, Clerk/WorkOS usually hurt less."},"lastReviewed":"2026-09-14","status":"active"},{"id":"stripe","slug":"stripe","name":"Stripe","website":"https://stripe.com","tagline":"Non-negotiable for most commercial agent products. Use Billing meters for token pass-through pricing.","description":"Non-negotiable for most commercial agent products. Use Billing meters for token pass-through pricing. Strengths: Default payments/billing API; Excellent docs and Agent/tooling ecosystem; Billing + metering for AI products.","categoryIds":["cat-auth-payments"],"jtbdIds":["payments","billing","usage-metering"],"regions":["global"],"pricingNotes":"Pay-as-you-go payment processing + Billing subscriptions; Connect for marketplaces; Atlas etc.","strengths":["Default payments/billing API","Excellent docs and Agent/tooling ecosystem","Billing + metering for AI products"],"weaknesses":["Fees add up globally","Account review risk","Not a full auth solution"],"tags":["payments","billing","connect","metering"],"scores":{"capabilityFit":9.6,"pricePerformance":7.8,"latencyReliability":9.4,"developerDx":9.6,"privacyPosture":8.5,"lockInRisk":6,"maturity":9.8,"overall":8.76,"kind":"curated_estimate","confidence":"high","asOf":"2026-09-14","sources":["Stripe: https://stripe.com","Stripe Billing: https://stripe.com/billing"],"notes":"Non-negotiable for most commercial agent products. Use Billing meters for token pass-through pricing."},"lastReviewed":"2026-09-14","status":"active"},{"id":"workos","slug":"workos","name":"WorkOS","website":"https://workos.com","tagline":"When your first big customer asks for Okta SSO tomorrow, WorkOS is how you say yes without a rewrite.","description":"When your first big customer asks for Okta SSO tomorrow, WorkOS is how you say yes without a rewrite. Strengths: Best modern enterprise SSO DX; Directory Sync that works; Complements Clerk/Auth0 rather than always replacing.","categoryIds":["cat-auth-payments"],"jtbdIds":["enterprise-sso","auth"],"regions":["global"],"pricingNotes":"Enterprise SSO/Directory Sync/Audit Log APIs; per-connection and MAU-style packaging.","strengths":["Best modern enterprise SSO DX","Directory Sync that works","Complements Clerk/Auth0 rather than always replacing"],"weaknesses":["Focused scope — not full consumer auth UI suite","Cost per enterprise connection","Still need user table somewhere"],"tags":["sso","scim","enterprise","directory-sync"],"scores":{"capabilityFit":9.2,"pricePerformance":7.5,"latencyReliability":9,"developerDx":9.4,"privacyPosture":8.8,"lockInRisk":4.5,"maturity":8.5,"overall":8.53,"kind":"curated_estimate","confidence":"high","asOf":"2026-09-14","sources":["WorkOS: https://workos.com"],"notes":"When your first big customer asks for Okta SSO tomorrow, WorkOS is how you say yes without a rewrite."},"lastReviewed":"2026-09-14","status":"active"},{"id":"midjourney","slug":"midjourney","name":"Midjourney","website":"https://www.midjourney.com","tagline":"Best looks, worst agent API story. For programmatic image agents prefer Fal/OpenAI/Stability unless you have…","description":"Best looks, worst agent API story. For programmatic image agents prefer Fal/OpenAI/Stability unless you have partner API access. Strengths: Top-tier aesthetic image quality; Huge cultural mindshare; Fast creative iteration in-product.","categoryIds":["cat-media-gen"],"jtbdIds":["image-generation"],"regions":["global"],"pricingNotes":"As of 2026-09-14: Primarily Discord/web subscription product; official public developer API historically limited/partner-gated. Treat programmatic access as constrained vs OpenAI/Fal/Stability. Subscription tiers for human creative use.","strengths":["Top-tier aesthetic image quality","Huge cultural mindshare","Fast creative iteration in-product"],"weaknesses":["API access not a first-class indie path historically","Automation/agent integration friction","ToS and commercial-use nuances"],"tags":["image","generative","discord","api-limited"],"scores":{"capabilityFit":9.4,"pricePerformance":7,"latencyReliability":8.5,"developerDx":4.5,"privacyPosture":6.5,"lockInRisk":7.5,"maturity":9,"overall":7.29,"kind":"curated_estimate","confidence":"medium","asOf":"2026-09-14","sources":["Midjourney: https://www.midjourney.com","Midjourney docs: https://docs.midjourney.com"],"notes":"Best looks, worst agent API story. For programmatic image agents prefer Fal/OpenAI/Stability unless you have partner API access."},"lastReviewed":"2026-09-14","status":"active"},{"id":"stability","slug":"stability","name":"Stability AI","website":"https://stability.ai","tagline":"Use API for convenience or self-host SD/Flux on RunPod/Fal for control. Great escape hatch from closed image…","description":"Use API for convenience or self-host SD/Flux on RunPod/Fal for control. Great escape hatch from closed image APIs. Strengths: Open weights + API; Customizable pipelines; Strong ecosystem.","categoryIds":["cat-media-gen"],"jtbdIds":["image-generation"],"regions":["global"],"pricingNotes":"As of 2026-09-14: Stable Diffusion / Stability platform APIs with credit-based billing; open weights freely self-hostable. Platform.stability.ai for hosted; escape hatch is self-host on RunPod/Fal/Modal.","strengths":["Open weights + API","Customizable pipelines","Strong ecosystem"],"weaknesses":["Company/API packaging has been turbulent historically","Quality vs Midjourney/Flux varies by model","Support perception mixed"],"tags":["image","sd","open-weights"],"scores":{"capabilityFit":8.6,"pricePerformance":8.5,"latencyReliability":7.5,"developerDx":7.8,"privacyPosture":8,"lockInRisk":2.5,"maturity":8.2,"overall":8.12,"kind":"curated_estimate","confidence":"high","asOf":"2026-09-14","sources":["Stability AI: https://stability.ai","Stability Developer Platform: https://platform.stability.ai","Stability pricing: https://platform.stability.ai/pricing"],"notes":"Use API for convenience or self-host SD/Flux on RunPod/Fal for control. Great escape hatch from closed image APIs."},"lastReviewed":"2026-09-14","status":"active"},{"id":"fal","slug":"fal","name":"Fal","website":"https://fal.ai","tagline":"Depot default for programmatic image/video/audio gen. One API to rule the media zoo.","description":"Depot default for programmatic image/video/audio gen. One API to rule the media zoo. Strengths: Best unified media model API for agents; Huge model catalog; Fast serverless DX.","categoryIds":["cat-media-gen","cat-inference-apis"],"jtbdIds":["image-generation","video-generation","media-generation","serverless-gpu","3d-generation"],"regions":["global"],"pricingNotes":"Serverless generative media API (1000+ models); per-second/per-output; extremely agent-friendly.","strengths":["Best unified media model API for agents","Huge model catalog","Fast serverless DX"],"weaknesses":["Costs climb with video","Abstraction over many upstream models","Less 'studio brand' than Runway/MJ"],"tags":["image","video","audio","serverless","models"],"scores":{"capabilityFit":9.2,"pricePerformance":8.5,"latencyReliability":8.6,"developerDx":9.6,"privacyPosture":7.2,"lockInRisk":4,"maturity":8,"overall":8.5,"kind":"curated_estimate","confidence":"high","asOf":"2026-10-19","sources":["Fal: https://fal.ai"],"notes":"Depot default for programmatic image/video/audio gen. One API to rule the media zoo."},"lastReviewed":"2026-10-19","status":"active"},{"id":"luma","slug":"luma","name":"Luma AI","website":"https://lumalabs.ai","tagline":"Solid video gen option — A/B with Runway/Pika/Fal-hosted models for your domain.","description":"Solid video gen option — A/B with Runway/Pika/Fal-hosted models for your domain. Strengths: Strong video generation quality; API available for products; 3D/NeRF heritage.","categoryIds":["cat-media-gen"],"jtbdIds":["video-generation","3d-generation"],"regions":["global"],"pricingNotes":"As of 2026-09-14: Dream Machine video and multimodal APIs with credit/subscription packaging plus builder API. Compare unit economics vs Runway/Pika/Fal-hosted video models for your domain.","strengths":["Strong video generation quality","API available for products","3D/NeRF heritage"],"weaknesses":["Video gen still expensive/slow vs images","Rapidly shifting competitive field","Rate limits on popular models"],"tags":["video","dream-machine","3d"],"scores":{"capabilityFit":8.8,"pricePerformance":7,"latencyReliability":8,"developerDx":8,"privacyPosture":7,"lockInRisk":5,"maturity":7.8,"overall":7.69,"kind":"curated_estimate","confidence":"medium","asOf":"2026-09-14","sources":["Luma AI: https://lumalabs.ai","Luma Dream Machine: https://lumalabs.ai/dream-machine","Luma API: https://docs.lumalabs.ai"],"notes":"Solid video gen option — A/B with Runway/Pika/Fal-hosted models for your domain."},"lastReviewed":"2026-09-14","status":"active"},{"id":"elevenlabs","slug":"elevenlabs","name":"ElevenLabs","website":"https://elevenlabs.io","tagline":"Default voice layer for talking agents. Combine with Twilio for phone; watch cloning policy.","description":"Default voice layer for talking agents. Combine with Twilio for phone; watch cloning policy. Strengths: Best-in-class TTS quality for many languages; Voice cloning + agents features; Good API for products.","categoryIds":["cat-media-gen","cat-comms"],"jtbdIds":["speech-synthesis","voice-agents","voice","speech-to-text","realtime-tts","voice-agent-platform"],"regions":["global"],"pricingNotes":"TTS/STS/voice agents; character/credit plans; telephony integrations (e.g. Twilio).","strengths":["Best-in-class TTS quality for many languages","Voice cloning + agents features","Good API for products"],"weaknesses":["Voice cloning ethics/ToS care required","Cost at high character volume","Alternatives catching up"],"tags":["tts","voice","audio","agents"],"scores":{"capabilityFit":9.4,"pricePerformance":7.5,"latencyReliability":8.8,"developerDx":9,"privacyPosture":7,"lockInRisk":5.5,"maturity":8.8,"overall":8.27,"kind":"curated_estimate","confidence":"high","asOf":"2026-09-14","sources":["ElevenLabs: https://elevenlabs.io"],"notes":"Default voice layer for talking agents. Combine with Twilio for phone; watch cloning policy."},"lastReviewed":"2026-09-14","status":"active"},{"id":"openai-media","slug":"openai-media","name":"OpenAI Media (Images/Audio)","website":"https://platform.openai.com/docs/guides/images","tagline":"Convenient if you're already on OpenAI. For best image variety/cost, keep Fal in the toolkit.","description":"Convenient if you're already on OpenAI. For best image variety/cost, keep Fal in the toolkit. Strengths: One vendor with chat+media; Solid quality and DX; Realtime/voice stack improving.","categoryIds":["cat-media-gen","cat-inference-apis"],"jtbdIds":["image-generation","speech-synthesis","speech-to-text","media-generation"],"regions":["global","us"],"pricingNotes":"Image generation, TTS, Whisper/transcription via OpenAI API; billed per image/minute/token as applicable.","strengths":["One vendor with chat+media","Solid quality and DX","Realtime/voice stack improving"],"weaknesses":["Not always best aesthetic vs Midjourney/Fal specialists","Pricing premium","Policy restrictions"],"tags":["image","tts","stt","openai"],"scores":{"capabilityFit":8.8,"pricePerformance":7.2,"latencyReliability":8.8,"developerDx":9.2,"privacyPosture":7.8,"lockInRisk":7,"maturity":9.6,"overall":8.15,"kind":"curated_estimate","confidence":"high","asOf":"2026-09-14","sources":["OpenAI Images: https://platform.openai.com/docs/guides/images","OpenAI Audio: https://platform.openai.com/docs/guides/audio"],"notes":"Convenient if you're already on OpenAI. For best image variety/cost, keep Fal in the toolkit."},"lastReviewed":"2026-09-14","status":"active"},{"id":"runway","slug":"runway","name":"Runway","website":"https://runwayml.com","tagline":"Top creative video brand. For pure agent pipelines, also price-check Fal-hosted alternatives.","description":"Top creative video brand. For pure agent pipelines, also price-check Fal-hosted alternatives. Strengths: Leading creative video tooling; API for integration; Strong brand in film/design.","categoryIds":["cat-media-gen"],"jtbdIds":["video-generation"],"regions":["global"],"pricingNotes":"Gen video models (Gen-3/etc.) via web + API; credit plans; creative-suite positioning.","strengths":["Leading creative video tooling","API for integration","Strong brand in film/design"],"weaknesses":["Credits get expensive","API/agent DX second to creative suite","Competition from Luma/Pika/Fal hosts"],"tags":["video","gen","creative"],"scores":{"capabilityFit":9,"pricePerformance":6.5,"latencyReliability":8.4,"developerDx":7.8,"privacyPosture":7,"lockInRisk":5.5,"maturity":8.8,"overall":7.75,"kind":"curated_estimate","confidence":"high","asOf":"2026-09-14","sources":["Runway: https://runwayml.com"],"notes":"Top creative video brand. For pure agent pipelines, also price-check Fal-hosted alternatives."},"lastReviewed":"2026-09-14","status":"active"},{"id":"pika","slug":"pika","name":"Pika API (API Club)","website":"https://dev.pika.art","tagline":"API story is no longer a mystery: self-serve keys + catalog quotes. Treat as a media aggregator (multi-vendor…","description":"API story is no longer a mystery: self-serve keys + catalog quotes. Treat as a media aggregator (multi-vendor models) competing with Fal/Replicate more than as “Pika-only video.” Prefer Fal for single-vendor media routing unless Club wholesale rates win your SKU mix. Strengths: Confirmed developer API with catalog, quotes, webhooks; Multi-model Gen Media aggregator pricing transparency; Agent-friendly docs (llms.txt / onboarding prompt).","categoryIds":["cat-media-gen"],"jtbdIds":["video-generation","media-generation","image-generation"],"regions":["global"],"pricingNotes":"As of 2026-09-14: Programmatic Gen Media API is GA via Pika API Club — $10/mo membership unlocks member/wholesale rates + $10 first-month usage credit; prepaid balance (micro-USD); pay-as-you-go per model from live catalog. 100+ video/image/audio/LLM ops via one key (api.dev.pika.art). No free generation tier. Enterprise: volume discounts, concurrency, reserved capacity.","strengths":["Confirmed developer API with catalog, quotes, webhooks","Multi-model Gen Media aggregator pricing transparency","Agent-friendly docs (llms.txt / onboarding prompt)"],"weaknesses":["Membership + prepaid friction vs pure PAYG hosts","Async-only — not realtime streaming media","Quality depends on upstream vendor per SKU"],"tags":["video","image","gen-media","api-aggregator"],"scores":{"capabilityFit":8.4,"pricePerformance":8.2,"latencyReliability":7.8,"developerDx":8.6,"privacyPosture":7,"lockInRisk":4.8,"maturity":7.8,"overall":7.88,"kind":"curated_estimate","confidence":"high","asOf":"2026-09-14","sources":["Pika API docs: https://dev.pika.art/","Pika API pricing: https://dev.pika.art/pricing","Pika API Club announcement: https://experiment.pika.art/blog/pika-api-club","Enterprise: https://dev.pika.art/enterprise"],"notes":"API story is no longer a mystery: self-serve keys + catalog quotes. Treat as a media aggregator (multi-vendor models) competing with Fal/Replicate more than as “Pika-only video.” Prefer Fal for single-vendor media routing unless Club wholesale rates win your SKU mix."},"lastReviewed":"2026-09-14","status":"active"},{"id":"tavily","slug":"tavily","name":"Tavily","website":"https://tavily.com","tagline":"Depot co-default for agent web search with Exa. Pick Tavily for classic SERP-like agent search.","description":"Depot co-default for agent web search with Exa. Pick Tavily for classic SERP-like agent search. Strengths: Agent-first search API; Simple integration with LangChain etc.; Optimized result payloads.","categoryIds":["cat-search-scrape"],"jtbdIds":["agent-search","search-augmented-llm","web-grounded-qa","neural-search","privacy-search","enterprise-search"],"regions":["global"],"pricingNotes":"Search API built for AI agents; credit-based; returns clean results for RAG.","strengths":["Agent-first search API","Simple integration with LangChain etc.","Optimized result payloads"],"weaknesses":["Not a full crawl platform","Coverage depends on upstream index","Younger than Bing/Brave"],"tags":["search","agents","rag"],"scores":{"capabilityFit":8.8,"pricePerformance":8.5,"latencyReliability":8.4,"developerDx":9.2,"privacyPosture":7.5,"lockInRisk":4.5,"maturity":7.8,"overall":8.29,"kind":"curated_estimate","confidence":"high","asOf":"2026-09-14","sources":["Tavily: https://tavily.com"],"notes":"Depot co-default for agent web search with Exa. Pick Tavily for classic SERP-like agent search."},"lastReviewed":"2026-09-14","status":"active"},{"id":"exa","slug":"exa","name":"Exa","website":"https://exa.ai","tagline":"Depot pick for research-style agent search. Pair with Firecrawl when you need full page content.","description":"Depot pick for research-style agent search. Pair with Firecrawl when you need full page content. Strengths: Semantic/neural search quality; Excellent for research agents; Clean developer API.","categoryIds":["cat-search-scrape"],"jtbdIds":["agent-search","neural-search","web-grounded-qa","llm-ready-extract","search-augmented-llm","privacy-search"],"regions":["global"],"pricingNotes":"Neural/embeddings-oriented web search API; usage-based; great for research agents.","strengths":["Semantic/neural search quality","Excellent for research agents","Clean developer API"],"weaknesses":["Different recall profile than keyword SERP","Cost at high volume","Complement not always replace keyword search"],"tags":["search","neural","research","agents"],"scores":{"capabilityFit":9,"pricePerformance":8.2,"latencyReliability":8.4,"developerDx":9.2,"privacyPosture":7.5,"lockInRisk":4.5,"maturity":8,"overall":8.3,"kind":"curated_estimate","confidence":"high","asOf":"2026-10-19","sources":["Exa: https://exa.ai"],"notes":"Depot pick for research-style agent search. Pair with Firecrawl when you need full page content."},"lastReviewed":"2026-10-19","status":"active"},{"id":"serper","slug":"serper","name":"Serper","website":"https://serper.dev","tagline":"Budget SERP hammer. Great for hacks; prefer Tavily/Exa for cleaner long-term agent search.","description":"Budget SERP hammer. Great for hacks; prefer Tavily/Exa for cleaner long-term agent search. Strengths: Cheap Google-like SERP; Simple JSON; Fast to integrate.","categoryIds":["cat-search-scrape"],"jtbdIds":["agent-search","serp-scrape"],"regions":["global"],"pricingNotes":"As of 2026-09-14: Low-cost Google SERP API; pay-per-query JSON results with generous free starter credits historically. Excellent budget hammer; prefer Tavily/Exa for LLM-shaped agent search long-term.","strengths":["Cheap Google-like SERP","Simple JSON","Fast to integrate"],"weaknesses":["Depends on scraping Google results — ToS/fragility risk","Less 'AI-native' packaging","Enterprise procurement may balk"],"tags":["search","serp","google","budget"],"scores":{"capabilityFit":8,"pricePerformance":9.5,"latencyReliability":8,"developerDx":8.8,"privacyPosture":6,"lockInRisk":4,"maturity":7.8,"overall":8.03,"kind":"curated_estimate","confidence":"high","asOf":"2026-09-14","sources":["Serper: https://serper.dev","Serper docs: https://serper.dev/docs"],"notes":"Budget SERP hammer. Great for hacks; prefer Tavily/Exa for cleaner long-term agent search."},"lastReviewed":"2026-09-14","status":"active"},{"id":"bing-search","slug":"bing-search","name":"Bing Search API (Azure)","website":"https://www.microsoft.com/en-us/bing/apis/bing-web-search-api","tagline":"Retired 2025-08-11 (Microsoft). Kept for history; not a candidate.","description":"Retired by Microsoft on 2025-08-11 (https://learn.microsoft.com/en-us/lifecycle/announcements/bing-search-api-retirement; re-checked 2026-10-21). Removed from all pick and runner-up lists on 2026-10-21; record and history kept. Former description: Enterprise-safe web search API. Indie agents often prefer Tavily/Exa for DX. Strengths: Official search API with enterprise terms; Azure billing/compliance; Stable documented API.","categoryIds":["cat-search-scrape"],"jtbdIds":["agent-search","enterprise-search","serp-scrape"],"regions":["global"],"pricingNotes":"Retired 2025-08-11 per Microsoft; no longer sold. Former note: Azure Cognitive Services Bing Web Search; per-transaction tiers; enterprise terms.","strengths":["Official search API with enterprise terms","Azure billing/compliance","Stable documented API"],"weaknesses":["Microsoft account/Azure friction","Result quality debates vs Google","Pricing less 'indie cute'"],"tags":["search","bing","azure","enterprise","sunset","retired"],"scores":{"capabilityFit":8.5,"pricePerformance":7.5,"latencyReliability":9,"developerDx":7.5,"privacyPosture":8.5,"lockInRisk":5.5,"maturity":9.5,"overall":8.07,"kind":"curated_estimate","confidence":"high","asOf":"2026-09-14","sources":["Bing Web Search API: https://www.microsoft.com/en-us/bing/apis/bing-web-search-api"],"notes":"Enterprise-safe web search API. Indie agents often prefer Tavily/Exa for DX."},"lastReviewed":"2026-09-14","status":"deprecated"},{"id":"brave-search","slug":"brave-search","name":"Brave Search API","website":"https://brave.com/search/api/","tagline":"Best privacy-forward search API with an independent index. Strong complement to Tavily/Exa.","description":"Best privacy-forward search API with an independent index. Strong complement to Tavily/Exa. Strengths: Own index (not just Google scrape); Privacy brand alignment; Straightforward HTTP API.","categoryIds":["cat-search-scrape"],"jtbdIds":["agent-search","privacy-search","serp-scrape","neural-search","enterprise-search"],"regions":["global"],"pricingNotes":"Independent index search API; plan-based request quotas; privacy-oriented positioning.","strengths":["Own index (not just Google scrape)","Privacy brand alignment","Straightforward HTTP API"],"weaknesses":["Index coverage vs Google still debated","Less agent-specific packaging than Tavily","Newer AI integrations"],"tags":["search","privacy","independent-index"],"scores":{"capabilityFit":8.2,"pricePerformance":8.5,"latencyReliability":8.6,"developerDx":8.2,"privacyPosture":9.2,"lockInRisk":3.5,"maturity":8.5,"overall":8.33,"kind":"curated_estimate","confidence":"high","asOf":"2026-09-14","sources":["Brave Search API: https://brave.com/search/api/"],"notes":"Best privacy-forward search API with an independent index. Strong complement to Tavily/Exa."},"lastReviewed":"2026-09-14","status":"active"},{"id":"baseten","slug":"baseten","name":"Baseten","website":"https://www.baseten.co","tagline":"Strong custom-model serving alternative to Replicate/Modal when you bring your own model.","description":"Strong custom-model serving alternative to Replicate/Modal when you bring your own model. Strengths: Production model serving focus; Good DX for custom models; Autoscaling inference.","categoryIds":["cat-gpu-cloud","cat-inference-apis"],"jtbdIds":["serverless-gpu","production-inference-fleet"],"regions":["us","global"],"pricingNotes":"As of 2026-09-14 (baseten.co/pricing live): Model APIs + dedicated Truss deploys. Token examples: GLM-5.3-Flash $0.15/$0.50 per 1M (stored row). Per baseten.co/pricing as of 2026-10-21 (not stored as rows): GLM-5.3 and GLM-5.2 each input $1.40, cache input $0.14, output $4.40 per 1M; GLM-5.3 Fast and GLM-5.2 Fast each input $2.10, cache input $0.21, output $6.60 per 1M. GLM 4.7 is no longer listed on that page as of 2026-10-21, so its figure was removed. Dedicated GPU/VM rates published on page (e.g. 16 GiB VM ~$0.01052, L4 ~$0.01414, A10G ~$0.02012, A100/B200 class higher). Basic self-serve; Pro priority GPUs; Enterprise BYOC/self-host. SOC2 Type II + HIPAA. No idle charge when scaled to zero.","strengths":["Production model serving focus","Good DX for custom models","Autoscaling inference"],"weaknesses":["Less model zoo than Replicate","Pricing vs Fireworks tradeoffs","Smaller community"],"tags":["inference","serving","truss","gpu"],"scores":{"capabilityFit":8.4,"pricePerformance":7.8,"latencyReliability":8.4,"developerDx":8.6,"privacyPosture":7.5,"lockInRisk":4.5,"maturity":8,"overall":7.99,"kind":"curated_estimate","confidence":"high","asOf":"2026-09-14","sources":["Baseten: https://www.baseten.co","Baseten docs: https://docs.baseten.co","Baseten pricing: https://www.baseten.co/pricing"],"notes":"Strong custom-model serving alternative to Replicate/Modal when you bring your own model."},"lastReviewed":"2026-09-14","status":"active"},{"id":"neon","slug":"neon","name":"Neon","website":"https://neon.tech","tagline":"Excellent serverless Postgres for agent apps that want branching + pgvector without Supabase's full BaaS.","description":"Excellent serverless Postgres for agent apps that want branching + pgvector without Supabase's full BaaS. Strengths: Serverless Postgres DX; Branching for previews/agents; pgvector without managing Postgres.","categoryIds":["cat-vector-memory","cat-object-storage"],"jtbdIds":["postgres-vector","serverless-db","app-backend"],"regions":["global","us","eu"],"pricingNotes":"Serverless Postgres with branching; pgvector supported; usage-based compute/storage.","strengths":["Serverless Postgres DX","Branching for previews/agents","pgvector without managing Postgres"],"weaknesses":["Not a dedicated vector DB","Cold starts on free compute","Scale limits vs dedicated vector engines"],"tags":["postgres","serverless","pgvector","branching"],"scores":{"capabilityFit":8.6,"pricePerformance":8.8,"latencyReliability":8.6,"developerDx":9.4,"privacyPosture":8.2,"lockInRisk":3,"maturity":8.5,"overall":8.59,"kind":"curated_estimate","confidence":"high","asOf":"2026-09-14","sources":["Neon: https://neon.tech"],"notes":"Excellent serverless Postgres for agent apps that want branching + pgvector without Supabase's full BaaS."},"lastReviewed":"2026-09-14","status":"active"},{"id":"s3","slug":"s3","name":"Amazon S3","website":"https://aws.amazon.com/s3/","tagline":"Default object store gravity well. Consider R2/Cloudflare or DO Spaces when egress dominates.","description":"Default object store gravity well. Consider R2/Cloudflare or DO Spaces when egress dominates. Strengths: Universal compatibility; Eleven nines durability narrative; Ecosystem gravity.","categoryIds":["cat-object-storage"],"jtbdIds":["object-storage"],"regions":["global"],"pricingNotes":"Industry-standard object storage; pay for storage/requests/egress; many tiers.","strengths":["Universal compatibility","Eleven nines durability narrative","Ecosystem gravity"],"weaknesses":["Egress fees painful","AWS account complexity","Not agent-specific"],"tags":["s3","object-storage","aws"],"scores":{"capabilityFit":9.8,"pricePerformance":7,"latencyReliability":9.8,"developerDx":8,"privacyPosture":8.8,"lockInRisk":5,"maturity":9.9,"overall":8.6,"kind":"curated_estimate","confidence":"high","asOf":"2026-09-14","sources":["Amazon S3: https://aws.amazon.com/s3/"],"notes":"Default object store gravity well. Consider R2/Cloudflare or DO Spaces when egress dominates."},"lastReviewed":"2026-09-14","status":"active"},{"id":"r2","slug":"r2","name":"Cloudflare R2","website":"https://developers.cloudflare.com/r2/","tagline":"Depot pick when agent artifacts get downloaded a lot. S3-compatible without the egress grief.","description":"Depot pick when agent artifacts get downloaded a lot. S3-compatible without the egress grief. Strengths: S3 API without egress tax; Works with Workers; Simple global story.","categoryIds":["cat-object-storage"],"jtbdIds":["object-storage","cheap-egress-storage"],"regions":["global"],"pricingNotes":"S3-compatible object storage with zero egress fees to the internet (Cloudflare positioning); storage+ops pricing.","strengths":["S3 API without egress tax","Works with Workers","Simple global story"],"weaknesses":["Feature parity gaps vs S3 occasionally","Vendor coupling to CF ecosystem","Consistency model nuances"],"tags":["s3-compat","object-storage","egress","cloudflare"],"scores":{"capabilityFit":9,"pricePerformance":9.5,"latencyReliability":9.2,"developerDx":8.8,"privacyPosture":8.2,"lockInRisk":4.5,"maturity":8.8,"overall":8.75,"kind":"curated_estimate","confidence":"high","asOf":"2026-10-19","sources":["Cloudflare R2: https://developers.cloudflare.com/r2/"],"notes":"Depot pick when agent artifacts get downloaded a lot. S3-compatible without the egress grief."},"lastReviewed":"2026-10-19","status":"active"},{"id":"novita","slug":"novita","name":"Novita AI","website":"https://novita.ai","tagline":"Budget open-model host with a pinned official matrix — strong DeepInfra/SiliconFlow peer. Prefer…","description":"Budget open-model host with a pinned official matrix — strong DeepInfra/SiliconFlow peer. Prefer Fireworks/Together when you need tighter SLAs or LoRA serving; Novita when sticker $/1M wins. Strengths: Aggressive open-model pricing; OpenAI-compatible API; Multi-modal catalog.","categoryIds":["cat-inference-apis","cat-gpu-cloud"],"jtbdIds":["cheap-volume-llm","open-model-inference","serverless-gpu","budget-open-inference"],"regions":["global","us"],"pricingNotes":"As of 2026-09-14 (novita.ai/model-api/pricing official): Full public token matrix — e.g. GPT-OSS-120B $0.05/$0.25 per 1M; DeepSeek V3.2 $0.269/$0.40; Llama 3.3 70B $0.135/$0.40; Qwen3 Coder 30B $0.07/$0.27; batch inference 50% intro discount. Also GPU instances + Agent Sandbox under same account. Multimodal (image/video/audio/search) separately metered.","strengths":["Aggressive open-model pricing","OpenAI-compatible API","Multi-modal catalog"],"weaknesses":["Less production brand gravity than Fireworks/Together","Capacity/quality variance by model","Newer ecosystem"],"tags":["llm","open-model","openai-compat","budget"],"scores":{"capabilityFit":7.2,"pricePerformance":9,"latencyReliability":7,"developerDx":7.8,"privacyPosture":7,"lockInRisk":3.5,"maturity":6.2,"overall":7.42,"kind":"curated_estimate","confidence":"high","asOf":"2026-09-14","sources":["Novita Model API pricing: https://novita.ai/model-api/pricing","Novita AI: https://novita.ai"],"notes":"Budget open-model host with a pinned official matrix — strong DeepInfra/SiliconFlow peer. Prefer Fireworks/Together when you need tighter SLAs or LoRA serving; Novita when sticker $/1M wins."},"lastReviewed":"2026-09-14","status":"active"},{"id":"hyperbolic","slug":"hyperbolic","name":"Hyperbolic","website":"https://www.hyperbolic.ai","tagline":"Cheap open-model inference peer to Novita/DeepInfra. Good for evals and non-critical volume.","description":"Cheap open-model inference peer to Novita/DeepInfra. Good for evals and non-critical volume. Strengths: Low open-model token prices; OpenAI SDK drop-in; Simple signup.","categoryIds":["cat-inference-apis"],"jtbdIds":["cheap-volume-llm","open-model-inference","budget-open-inference"],"regions":["global","us"],"pricingNotes":"As of 2026-09-14 (docs.hyperbolic.ai performance-limits): PAYG no monthly commitment. Text: small (3B–8B) from **$0.10/1M**; medium (32B–72B) **$0.20–$0.40/1M**; large (120B–480B) **$0.30–$4.00/1M**. Image base $0.01 (1024²/25 steps). Audio $5/1M chars. Tiers: Basic 60 RPM free; Pro 600 RPM with **≥$5 deposit**; Enterprise custom. OpenAI-compatible api.hyperbolic.xyz/v1. Per-model card in Text APIs docs — re-curl before large commits.","strengths":["Low open-model token prices","OpenAI SDK drop-in","Simple signup"],"weaknesses":["Smaller catalog than Together/Fireworks","Less enterprise narrative","Rate limits on free tier"],"tags":["llm","open-model","openai-compat","budget"],"scores":{"capabilityFit":7,"pricePerformance":8.8,"latencyReliability":6.8,"developerDx":8,"privacyPosture":6.8,"lockInRisk":3.2,"maturity":5.8,"overall":7.29,"kind":"curated_estimate","confidence":"high","asOf":"2026-09-14","sources":["Hyperbolic performance & pricing: https://www.hyperbolic.ai/docs/inference/performance-limits","Hyperbolic billing: https://www.hyperbolic.ai/docs/general/billing-payments","Hyperbolic: https://www.hyperbolic.ai","Hyperbolic docs: https://docs.hyperbolic.ai","Inference price compare: https://induwara.lk/tools/ai-inference-provider-comparison"],"notes":"Cheap open-model inference peer to Novita/DeepInfra. Good for evals and non-critical volume."},"lastReviewed":"2026-09-14","status":"active"},{"id":"infermatic","slug":"infermatic","name":"Infermatic","website":"https://infermatic.ai","tagline":"Only choose if your steady RPM fits inside 12–18 and flat $9–$20 beats DeepInfra/Novita per-token math. Fine…","description":"Only choose if your steady RPM fits inside 12–18 and flat $9–$20 beats DeepInfra/Novita per-token math. Fine for hobby/roleplay throughput; not a production agent default. Strengths: Clear published flat-rate + RPM matrix; OpenAI-compatible API; crypto/card billing; No-log privacy marketing.","categoryIds":["cat-inference-apis"],"jtbdIds":["open-model-inference","cheap-volume-llm"],"regions":["global","us"],"pricingNotes":"As of 2026-09-14 (official infermatic.ai/pricing): Flat-rate plans — Free $0 (no API), Essential $9/mo (1 parallel, 12 RPM), Standard $16/mo (1 parallel, 15 RPM), Plus $20/mo (2 parallel, 18 RPM). OpenAI-compatible API on paid tiers; curated models through 72B+ / MoE on higher plans; no per-token usage fees. Privacy: marketed no-log of prompts/outputs.","strengths":["Clear published flat-rate + RPM matrix","OpenAI-compatible API; crypto/card billing","No-log privacy marketing"],"weaknesses":["Hard RPM ceilings (≤18) kill serious agents","Niche catalog vs DeepInfra/Fireworks","Low production credibility / SLA clarity"],"tags":["llm","open-model","niche"],"scores":{"capabilityFit":5.5,"pricePerformance":8,"latencyReliability":5.8,"developerDx":7,"privacyPosture":7.8,"lockInRisk":3.5,"maturity":5.5,"overall":6.52,"kind":"curated_estimate","confidence":"high","asOf":"2026-09-14","sources":["Infermatic pricing: https://infermatic.ai/pricing/","Infermatic docs: https://ui.infermatic.ai/docs","Infermatic: https://infermatic.ai/"],"notes":"Only choose if your steady RPM fits inside 12–18 and flat $9–$20 beats DeepInfra/Novita per-token math. Fine for hobby/roleplay throughput; not a production agent default."},"lastReviewed":"2026-09-14","status":"active"},{"id":"siliconflow","slug":"siliconflow","name":"SiliconFlow","website":"https://www.siliconflow.com","tagline":"Price/catalog monster for open models. Confirm data-residency/compliance before production.","description":"Price/catalog monster for open models. Confirm data-residency/compliance before production. Strengths: Huge open-model catalog; Aggressive pricing; OpenAI-compat.","categoryIds":["cat-inference-apis"],"jtbdIds":["cheap-volume-llm","open-model-inference","embeddings","budget-open-inference"],"regions":["global","us"],"pricingNotes":"As of 2026-09-14 (siliconflow.com/pricing live): PAYG serverless, $1 free credit, no commitments. Token examples (per 1M): DeepSeek-V4-Flash $0.13/$0.28; DeepSeek-V4-Pro ~$1.50/$3.14; gpt-oss-120b $0.05/$0.45; gpt-oss-20b $0.04/$0.18; GLM-5.3 $1.4/$4.4; GLM-5.3-Flash $0.15/$0.50; MiniMax-M2.5/M3 $0.3/$1.2; Kimi-K2.5 $0.45/$2.25; Kimi-K3 $2.7/$13.5; Qwen3.5-9B $0.1/$0.15. Images: FLUX.2 pro $0.03, flex $0.06, Z-Image-Turbo $0.005. Video Wan2.2 ~$0.29/video.","strengths":["Huge open-model catalog","Aggressive pricing","OpenAI-compat"],"weaknesses":["Less Western enterprise footprint","Verify latency/SLA for your region","Compliance diligence needed"],"tags":["llm","open-model","openai-compat","budget","china-adjacent"],"scores":{"capabilityFit":7.4,"pricePerformance":9.2,"latencyReliability":6.8,"developerDx":7.6,"privacyPosture":6.2,"lockInRisk":3.5,"maturity":6,"overall":7.34,"kind":"curated_estimate","confidence":"high","asOf":"2026-09-14","sources":["SiliconFlow: https://www.siliconflow.com","SiliconFlow pricing: https://www.siliconflow.com/pricing","Cheapest inference Jul 2026: https://infrabase.ai/blog/cheapest-inference-providers"],"notes":"Price/catalog monster for open models. Confirm data-residency/compliance before production."},"lastReviewed":"2026-09-14","status":"active"},{"id":"lepton","slug":"lepton","name":"Lepton AI","website":"https://www.lepton.ai","tagline":"Consider when Fireworks latency/price disappoints on a supported model. Keep a second host.","description":"Consider when Fireworks latency/price disappoints on a supported model. Keep a second host. Strengths: Low-latency serverless narrative; GPU + inference in one; OpenAI-compat.","categoryIds":["cat-inference-apis","cat-gpu-cloud"],"jtbdIds":["open-model-inference","ultra-low-latency","serverless-gpu"],"regions":["global","us"],"pricingNotes":"As of 2026-09-14: Serverless LLM inference + GPU cloud; OpenAI-compatible REST. Secondary quotes put serverless token rates competitive (often under Fireworks on some SKUs) and GPU examples like A10G from ~$0.75/hr. Smaller catalog than Together/Fireworks; reputation for low median latency on supported models.","strengths":["Low-latency serverless narrative","GPU + inference in one","OpenAI-compat"],"weaknesses":["Narrower catalog","Less mindshare than Fireworks/Together","Pricing pages less transparent"],"tags":["llm","open-model","serverless","latency"],"scores":{"capabilityFit":7.2,"pricePerformance":8.2,"latencyReliability":7.2,"developerDx":7.8,"privacyPosture":7,"lockInRisk":3.8,"maturity":6.4,"overall":7.3,"kind":"curated_estimate","confidence":"medium","asOf":"2026-09-14","sources":["Lepton AI: https://www.lepton.ai","Fireworks vs Lepton 2026: https://costbench.com/compare/fireworks-ai-vs-lepton/","Fireworks alternatives 2026: https://www.spheron.network/blog/fireworks-ai-alternatives/"],"notes":"Consider when Fireworks latency/price disappoints on a supported model. Keep a second host."},"lastReviewed":"2026-09-14","status":"active"},{"id":"anyscale","slug":"anyscale","name":"Anyscale","website":"https://www.anyscale.com","tagline":"Default when your agent stack already is Ray or you need serious multi-node Python. For simple serverless GPU…","description":"Default when your agent stack already is Ray or you need serious multi-node Python. For simple serverless GPU prefer Modal/RunPod. Strengths: Best-in-class Ray/distributed Python; Production multi-node jobs; Enterprise support.","categoryIds":["cat-gpu-cloud","cat-cpu-compute","cat-train-finetune"],"jtbdIds":["multi-node-training","production-inference-fleet","agent-backend","ray-distributed-agents"],"regions":["global","us"],"pricingNotes":"As of 2026-09-14 (anyscale.com/pricing): PAYG in Anyscale Credits, no monthly platform fee. Hosted examples: CPU AC 0.0135/hr; T4 0.5682; L4 0.9542; A10G 1.3635; A100 4.9591; H/B/GB families listed as contact. Hosted (Anyscale-managed, limited regions) vs BYOC (your VPC, use existing GPU reservations, marketplace billing). OSS Ray remains free; Anyscale is the managed control plane.","strengths":["Best-in-class Ray/distributed Python","Production multi-node jobs","Enterprise support"],"weaknesses":["Heavier than Modal for simple endpoints","Not a frontier LLM API","Sales motion for large accounts"],"tags":["ray","distributed","training","inference-fleet"],"scores":{"capabilityFit":8.2,"pricePerformance":6.5,"latencyReliability":8,"developerDx":7.5,"privacyPosture":7.8,"lockInRisk":5.5,"maturity":8.2,"overall":7.46,"kind":"curated_estimate","confidence":"high","asOf":"2026-09-14","sources":["Anyscale pricing: https://www.anyscale.com/pricing","Anyscale: https://www.anyscale.com","Ray: https://www.ray.io"],"notes":"Default when your agent stack already is Ray or you need serious multi-node Python. For simple serverless GPU prefer Modal/RunPod."},"lastReviewed":"2026-09-14","status":"active"},{"id":"databricks-mosaic","slug":"databricks-mosaic","name":"Databricks Mosaic AI","website":"https://www.databricks.com/product/machine-learning","tagline":"Pick only when your data/platform already is Databricks. Otherwise Bedrock/Vertex or direct APIs.","description":"Pick only when your data/platform already is Databricks. Otherwise Bedrock/Vertex or direct APIs. Strengths: Data gravity with lakehouse; Enterprise governance; Training + serving + evals in one platform.","categoryIds":["cat-inference-apis","cat-train-finetune","cat-vector-memory"],"jtbdIds":["enterprise-llm","finetune","rag-enterprise","lakehouse-llm","sql-warehouse-agent-context"],"regions":["global","us"],"pricingNotes":"As of 2026-09-14: Lakehouse-native model training/serving billed via Databricks DBUs + cloud compute. Foundation Model Serving and Mosaic tooling sit inside Databricks contracts — not indie per-token pricing. Strong if data already lives in Databricks.","strengths":["Data gravity with lakehouse","Enterprise governance","Training + serving + evals in one platform"],"weaknesses":["Expensive/complex for startups","Lock-in to Databricks","Overkill for simple agent APIs"],"tags":["enterprise","lakehouse","serving","training"],"scores":{"capabilityFit":8.4,"pricePerformance":4.8,"latencyReliability":8.5,"developerDx":6.8,"privacyPosture":8.2,"lockInRisk":7.8,"maturity":8.8,"overall":7.11,"kind":"curated_estimate","confidence":"medium","asOf":"2026-09-14","sources":["Databricks Mosaic AI: https://www.databricks.com/product/machine-learning","Databricks pricing: https://www.databricks.com/product/pricing","AI-ready data platforms 2026: https://www.nx1.io/blog/top-ai-ready-data-platforms-enterprise-multi-cloud-2026"],"notes":"Pick only when your data/platform already is Databricks. Otherwise Bedrock/Vertex or direct APIs."},"lastReviewed":"2026-09-14","status":"active"},{"id":"snowflake-cortex","slug":"snowflake-cortex","name":"Snowflake Cortex","website":"https://www.snowflake.com/en/data-cloud/cortex/","tagline":"Use when Snowflake already holds the data. Don't drag agent backends into Snowflake just for Cortex.","description":"Use when Snowflake already holds the data. Don't drag agent backends into Snowflake just for Cortex. Strengths: SQL-native AI on governed data; Enterprise residency/RBAC; No separate vector infra for many RAG patterns.","categoryIds":["cat-inference-apis","cat-vector-memory"],"jtbdIds":["enterprise-llm","rag-enterprise","embeddings","lakehouse-llm","sql-warehouse-agent-context"],"regions":["global","us"],"pricingNotes":"As of 2026-09-14: SQL-first AI functions (COMPLETE, EMBED, SEARCH, etc.) priced in Snowflake AI credits (secondary sources cite ~$2.00 per AI credit — confirm account rate card). Attractive when data already in Snowflake; not a general agent LLM default.","strengths":["SQL-native AI on governed data","Enterprise residency/RBAC","No separate vector infra for many RAG patterns"],"weaknesses":["Not agent-tooling native","Credit math opaque","Lock-in to Snowflake warehouse economics"],"tags":["enterprise","sql","rag","snowflake"],"scores":{"capabilityFit":7.8,"pricePerformance":5.5,"latencyReliability":8.4,"developerDx":7,"privacyPosture":8.5,"lockInRisk":8,"maturity":8.6,"overall":7.1,"kind":"curated_estimate","confidence":"medium","asOf":"2026-09-14","sources":["Snowflake Cortex: https://www.snowflake.com/en/data-cloud/cortex/","Cortex docs: https://docs.snowflake.com/en/user-guide/snowflake-cortex/overview","AI-ready data platforms 2026: https://www.nx1.io/blog/top-ai-ready-data-platforms-enterprise-multi-cloud-2026"],"notes":"Use when Snowflake already holds the data. Don't drag agent backends into Snowflake just for Cortex."},"lastReviewed":"2026-09-14","status":"active"},{"id":"cloudflare-workers-ai","slug":"cloudflare-workers-ai","name":"Cloudflare Workers AI","website":"https://developers.cloudflare.com/workers-ai/","tagline":"Best when your agent already runs on Workers and you want colocated small/open models. Not a frontier brain.","description":"Best when your agent already runs on Workers and you want colocated small/open models. Not a frontier brain. Strengths: Edge + Workers colocation; Transparent neuron/token pricing; Generous free daily allocation; Embeddings/image/audio too.","categoryIds":["cat-inference-apis","cat-cpu-compute"],"jtbdIds":["edge-deploy","cheap-volume-llm","embeddings","speech-to-text","edge-llm-inference","wasm-edge-small-models"],"regions":["global","us"],"pricingNotes":"As of 2026-10-24 (developers.cloudflare.com/workers-ai/platform/pricing, fetched 2026-10-24): '$0.011 per 1,000 Neurons' with '10,000 Neurons per day at no charge' on both Workers Free and Workers Paid ('All limits reset daily at 00:00 UTC.'). 'To use more than 10,000 Neurons per day, you need to sign up for the Workers Paid plan'. 'Some models require a paid billing method' (listed Kimi K2.6/K2.7-code, GLM-5.2/5.3 and DeepSeek V4 models), payable via Workers Paid or 'prepaid AI Gateway credits'. Per-model token prices are shown as equivalents while 'still billing in neurons in the back end': @cf/meta/llama-3.1-8b-instruct-fp8-fast $0.045/$0.384 per M input/output tokens ('$0.045 per M input tokens', '$0.384 per M output tokens'); @cf/openai/gpt-oss-120b $0.350/$0.750 per M input/output tokens ('$0.350 per M input tokens', '$0.750 per M output tokens'); @cf/baai/bge-m3 '$0.012 per M input tokens'.","strengths":["Edge + Workers colocation","Transparent neuron/token pricing","Generous free daily allocation","Embeddings/image/audio too"],"weaknesses":["Not frontier-quality vs Anthropic/OpenAI","Model list curated by CF","Some models require Paid plan"],"tags":["edge","llm","embeddings","workers","openai-compat"],"scores":{"capabilityFit":6.8,"pricePerformance":8.5,"latencyReliability":8.2,"developerDx":8.4,"privacyPosture":8,"lockInRisk":4.5,"maturity":7.8,"overall":7.69,"kind":"curated_estimate","confidence":"high","asOf":"2026-09-14","sources":["Workers AI pricing: https://developers.cloudflare.com/workers-ai/platform/pricing/","Workers AI: https://developers.cloudflare.com/workers-ai/","Cloudflare Workers AI product: https://www.cloudflare.com/products/workers-ai/"],"notes":"Best when your agent already runs on Workers and you want colocated small/open models. Not a frontier brain."},"lastReviewed":"2026-09-14","status":"active"},{"id":"nvidia-nim","slug":"nvidia-nim","name":"NVIDIA NIM / NGC","website":"https://www.nvidia.com/en-us/ai/","tagline":"Choose when you self-host inference on NVIDIA GPUs and want blessed containers. For serverless tokens use…","description":"Choose when you self-host inference on NVIDIA GPUs and want blessed containers. For serverless tokens use Fireworks/Groq. Strengths: Best CUDA-optimized serving stacks; Portable across GPU clouds; Enterprise support path.","categoryIds":["cat-inference-apis","cat-gpu-cloud","cat-train-finetune"],"jtbdIds":["production-inference-fleet","open-model-inference","enterprise-gpu","self-host-nim","air-gapped-model-serve"],"regions":["global","us"],"pricingNotes":"As of 2026-10-24 (nvidia.com/en-us/ai-data-science/products/nim-microservices, fetched 2026-10-24; NVIDIA publishes no NIM price page, and this product page states no price figures, so production licence price stays null — 'Contact Sales'): hosted prototyping 'Start Prototyping for Free' ('Access for development and testing as part of the NVIDIA Developer Program'); self-hosting 'Start with free access for development and testing, and move to an NVIDIA AI Enterprise license for production.' Deploy as 'hosted APIs for NIM accelerated by NVIDIA DGX™ Cloud' or 'self-hosting models on your preferred infrastructure'.","strengths":["Best CUDA-optimized serving stacks","Portable across GPU clouds","Enterprise support path"],"weaknesses":["Not a simple hosted LLM API","Ops burden vs Fireworks","License complexity"],"tags":["nvidia","nim","ngc","self-host","enterprise"],"scores":{"capabilityFit":8.6,"pricePerformance":7,"latencyReliability":8,"developerDx":6.8,"privacyPosture":8.5,"lockInRisk":6,"maturity":9,"overall":7.66,"kind":"curated_estimate","confidence":"medium","asOf":"2026-10-19","sources":["NVIDIA NIM: https://www.nvidia.com/en-us/ai/","NVIDIA NGC: https://catalog.ngc.nvidia.com","NIM docs: https://docs.nvidia.com/nim/"],"notes":"Choose when you self-host inference on NVIDIA GPUs and want blessed containers. For serverless tokens use Fireworks/Groq."},"lastReviewed":"2026-10-19","status":"active"},{"id":"nebius","slug":"nebius","name":"Nebius","website":"https://nebius.com","tagline":"Top-tier neocloud when you want list prices + serious clusters. RunPod/Vast still win for hackathon…","description":"Top-tier neocloud when you want list prices + serious clusters. RunPod/Vast still win for hackathon self-serve. Strengths: Transparent public GPU price list; Blackwell + Hopper SKUs; Full AI cloud stack; Commitment discounts.","categoryIds":["cat-gpu-cloud"],"jtbdIds":["predictable-gpu","enterprise-gpu","multi-node-training","neocloud-gpu"],"regions":["global","us"],"pricingNotes":"As of 2026-09-14 (post Jun 1 2026 list): H100 NVLink $3.85/hr on-demand / $2.15 preemptible; H200 $4.50/$2.45; B200 $7.15/$3.95; B300 $7.85/$4.30; RTX PRO 6000 $1.80/$0.95. Commitment discounts up to ~35%. Full AI cloud (K8s, Slurm, storage) not just bare GPUs.","strengths":["Transparent public GPU price list","Blackwell + Hopper SKUs","Full AI cloud stack","Commitment discounts"],"weaknesses":["Pricier than marketplace GPUs","Less indie DX than RunPod","EU/US capacity planning needed"],"tags":["gpu","h100","blackwell","neocloud"],"scores":{"capabilityFit":8.6,"pricePerformance":7,"latencyReliability":8.4,"developerDx":7.6,"privacyPosture":8,"lockInRisk":4.8,"maturity":7.8,"overall":7.75,"kind":"curated_estimate","confidence":"high","asOf":"2026-09-14","sources":["Nebius prices: https://nebius.com/prices","Nebius compute pricing docs: https://docs.nebius.com/compute/resources/pricing","Nebius vs Oblivus 2026: https://computeprices.com/compare/nebius-vs-oblivus"],"notes":"Top-tier neocloud when you want list prices + serious clusters. RunPod/Vast still win for hackathon self-serve."},"lastReviewed":"2026-09-14","status":"active"},{"id":"tensordock","slug":"tensordock","name":"TensorDock","website":"https://tensordock.com","tagline":"Cheap GPU peer to Vast/RunPod community. Fine for interruptible training/batch; not regulated PII.","description":"Cheap GPU peer to Vast/RunPod community. Fine for interruptible training/batch; not regulated PII. Strengths: Aggressive marketplace pricing; Wide consumer+datacenter SKUs; Per-second components.","categoryIds":["cat-gpu-cloud"],"jtbdIds":["cheap-gpu","batch-gpu","distributed-cheap-gpu"],"regions":["global","us"],"pricingNotes":"As of 2026-09-14: Marketplace GPU cloud; published examples H100 SXM5 from ~$2.25/hr, A100 SXM4 ~$1.80, RTX 4090 ~$0.35 (host-dependent). Per-second billing for CPU/RAM/disk add-ons. Host variability like other marketplaces.","strengths":["Aggressive marketplace pricing","Wide consumer+datacenter SKUs","Per-second components"],"weaknesses":["Host-dependent reliability","Security tenancy varies","Less SLA than CoreWeave/Nebius"],"tags":["gpu","marketplace","budget"],"scores":{"capabilityFit":7.2,"pricePerformance":8.8,"latencyReliability":6.2,"developerDx":7.2,"privacyPosture":5.5,"lockInRisk":3,"maturity":6.5,"overall":7.08,"kind":"curated_estimate","confidence":"medium","asOf":"2026-09-14","sources":["TensorDock GPUs: https://marketplace.tensordock.com/cloud-gpus.html","TensorDock: https://tensordock.com"],"notes":"Cheap GPU peer to Vast/RunPod community. Fine for interruptible training/batch; not regulated PII."},"lastReviewed":"2026-09-14","status":"active"},{"id":"massed-compute","slug":"massed-compute","name":"Massed Compute","website":"https://massedcompute.com","tagline":"Strong self-serve neocloud with a durable public rate card and owned fleet (not a marketplace). Good…","description":"Strong self-serve neocloud with a durable public rate card and owned fleet (not a marketplace). Good Nebius/RunPod peer when you want listed H100/H200 without shadeform aggregation. Still medium-maturity vs CoreWeave. Strengths: Transparent on-demand list; Own infra narrative; Bare metal + clusters.","categoryIds":["cat-gpu-cloud"],"jtbdIds":["cheap-gpu","predictable-gpu","batch-gpu"],"regions":["global","us"],"pricingNotes":"As of 2026-09-14 (vm.massedcompute.com/pricing official): Owns/operates infra, no bandwidth charges claimed. H100 80GB $2.73/hr (1×); H100 SXM5 $2.89/hr; H100 NVL $3.11/hr; H200 NVL $3.62/hr; A100 80GB $1.35/hr; L40S $0.97/hr; RTX 6000 Ada $0.79/hr; RTX PRO 6000 Blackwell $2.19/hr; RTX PRO 4500 Blackwell $0.92/hr; B200/B300 8× cluster SKUs published. Bare metal/IB clusters custom-quoted.","strengths":["Transparent on-demand list","Own infra narrative","Bare metal + clusters"],"weaknesses":["Smaller brand than CoreWeave/Lambda","Fewer regions","Sales for large clusters"],"tags":["gpu","h100","on-demand"],"scores":{"capabilityFit":7.6,"pricePerformance":8.2,"latencyReliability":7.4,"developerDx":7.2,"privacyPosture":7.5,"lockInRisk":4,"maturity":6.8,"overall":7.42,"kind":"curated_estimate","confidence":"high","asOf":"2026-09-14","sources":["Massed Compute on-demand GPU pricing: https://vm.massedcompute.com/pricing","Massed Compute: https://massedcompute.com"],"notes":"Strong self-serve neocloud with a durable public rate card and owned fleet (not a marketplace). Good Nebius/RunPod peer when you want listed H100/H200 without shadeform aggregation. Still medium-maturity vs CoreWeave."},"lastReviewed":"2026-09-14","status":"active"},{"id":"shadeform","slug":"shadeform","name":"Shadeform","website":"https://shadeform.com","tagline":"Use as a router/compare layer across GPU clouds. Still pin production to a known underlying provider.","description":"Use as a router/compare layer across GPU clouds. Still pin production to a known underlying provider. Strengths: One API across many GPU clouds; Live price/availability compare; Reduces vendor hopping friction.","categoryIds":["cat-gpu-cloud"],"jtbdIds":["cheap-gpu","predictable-gpu","gpu-marketplace-router"],"regions":["global","us"],"pricingNotes":"As of 2026-09-14: GPU marketplace aggregating 30+ clouds with unified API/billing; prices are provider-pass-through (not Shadeform's own SKUs). Useful meta-layer to compare Nebius/Oblivus/RunPod/etc. without N consoles.","strengths":["One API across many GPU clouds","Live price/availability compare","Reduces vendor hopping friction"],"weaknesses":["Meta-layer adds dependency","SLA is underlying provider's","Not capacity itself"],"tags":["gpu","marketplace","aggregator"],"scores":{"capabilityFit":7,"pricePerformance":8.4,"latencyReliability":6.8,"developerDx":8,"privacyPosture":6,"lockInRisk":4.5,"maturity":6.2,"overall":7.09,"kind":"curated_estimate","confidence":"medium","asOf":"2026-09-14","sources":["Shadeform: https://shadeform.com","Shadeform prices: https://shadeform.com/prices","Shadeform marketplace overview: https://runany.dev/blog/shadeform-gpu-cloud-marketplace/"],"notes":"Use as a router/compare layer across GPU clouds. Still pin production to a known underlying provider."},"lastReviewed":"2026-09-14","status":"active"},{"id":"salad","slug":"salad","name":"SaladCloud","website":"https://salad.com","tagline":"Promoted high on official /pricing floor ($0.015/GPU-hr). Depot interruptible ultra-cheap GPU pick…","description":"Promoted high on official /pricing floor ($0.015/GPU-hr). Depot interruptible ultra-cheap GPU pick; Vast/TensorDock remain marketplace peers. Prefer Nebius/RunPod/Modal when you need sticky sessions or predictable TTFT. Strengths: Extremely cheap GPUs; Huge distributed capacity; Container-native.","categoryIds":["cat-gpu-cloud"],"jtbdIds":["cheap-gpu","batch-gpu","distributed-cheap-gpu","on-device-edge-batch"],"regions":["global","us"],"pricingNotes":"As of 2026-10-24 (salad.com/pricing, fetched 2026-10-24): GPU classes 'from $0.015 per GPU-hour at Lowest priority'; 'Billing is per second while an instance is in the running state.' 'Allocation, image download, and container start are free.' GPU prices are 'Per GPU, per hour. vCPU and RAM included.' across four priority tiers (High, Medium, Low, Lowest), e.g. RTX 4090 24 GB '$0.330' High to '$0.160' Lowest; RTX 5090 32 GB '$0.500' High to '$0.250' Lowest. High: 'Never preempted by other workloads.' Medium and Low: 'May be preempted by higher-priority groups.' 'Lowest is the lowest cost and the most interruptible.' CPU-only: 'Billed at $0.005 per vCPU-hour and $0.001 per GB of RAM per hour.' (1 vCPU / 1 GB '$0.006' per hr). Volume discounts 'for deployments above 100 GPUs, long-running jobs, and committed contracts' (sales); Salad Dedicated RTX PRO 6000 Blackwell 'quoted per allocation and term'. Free trial: 'No.'","strengths":["Extremely cheap GPUs","Huge distributed capacity","Container-native"],"weaknesses":["Interruptible/priority variance","Consumer GPU mix","Not enterprise SLA"],"tags":["gpu","distributed","budget","batch"],"scores":{"capabilityFit":6.5,"pricePerformance":9.5,"latencyReliability":5,"developerDx":7,"privacyPosture":4.5,"lockInRisk":3.5,"maturity":6.8,"overall":6.72,"kind":"curated_estimate","confidence":"high","asOf":"2026-10-19","sources":["SaladCloud pricing: https://salad.com/pricing","Salad price changes Sep 2026: https://blog.salad.com/saladcloud-price-changes-september-2026/","SaladCloud: https://salad.com"],"notes":"Promoted high on official /pricing floor ($0.015/GPU-hr). Depot interruptible ultra-cheap GPU pick; Vast/TensorDock remain marketplace peers. Prefer Nebius/RunPod/Modal when you need sticky sessions or predictable TTFT."},"lastReviewed":"2026-10-19","status":"active"},{"id":"oblivus","slug":"oblivus","name":"Oblivus","website":"https://oblivus.com","tagline":"Price-check peer to Nebius/RunPod for on-demand H100s. Validate support and regions.","description":"Price-check peer to Nebius/RunPod for on-demand H100s. Validate support and regions. Strengths: Competitive H100/H200 quotes; Consumer + datacenter mix; Simple cloud UX.","categoryIds":["cat-gpu-cloud"],"jtbdIds":["cheap-gpu","predictable-gpu"],"regions":["global","us"],"pricingNotes":"As of 2026-09-14: GPU cloud with competitive list rates; secondary trackers quote H100 SXM ~$3.50/hr, H200 ~$3.99, A100 ~$1.47, RTX 4090 ~$0.64 (Aug 2026 snapshots). Compare live vs Nebius/RunPod before locking.","strengths":["Competitive H100/H200 quotes","Consumer + datacenter mix","Simple cloud UX"],"weaknesses":["Smaller ecosystem","Verify regions/SLA","Less known than CoreWeave"],"tags":["gpu","h100","neocloud"],"scores":{"capabilityFit":7.4,"pricePerformance":8.4,"latencyReliability":7,"developerDx":7.2,"privacyPosture":6.8,"lockInRisk":3.8,"maturity":6,"overall":7.21,"kind":"curated_estimate","confidence":"medium","asOf":"2026-09-14","sources":["Oblivus: https://oblivus.com","Nebius vs Oblivus 2026: https://computeprices.com/compare/nebius-vs-oblivus"],"notes":"Price-check peer to Nebius/RunPod for on-demand H100s. Validate support and regions."},"lastReviewed":"2026-09-14","status":"active"},{"id":"composio","slug":"composio","name":"Composio","website":"https://composio.dev","tagline":"Depot pick for tool/MCP router with managed OAuth across 1500+ apps. Official rate card is now public — Hobby…","description":"Depot pick for tool/MCP router with managed OAuth across 1500+ apps. Official rate card is now public — Hobby is real free for light agents; Pro is the agent-scale default. Not a compute fabric; pair with E2B/Browserbase when you need sandboxes/browsers. Strengths: Huge tool catalog + MCP; Managed OAuth; Fast agent integration.","categoryIds":["cat-sandboxes-browsers","cat-search-scrape"],"jtbdIds":["agent-tools","web-agent","agent-sandbox","agent-tool-router","calendar-scheduling-tools"],"regions":["global","us"],"pricingNotes":"As of 2026-10-24 (composio.dev/pricing, fetched 2026-10-24; 'New pricing applies to signups on or after August 15, 2026'): Hobby $0 — '100,000 tool calls, 50,000 triggers, and 3 team members included free each month'; 'The Hobby plan is hard-capped'. Pro '$29 / mo' with '$29 monthly usage credit included' ('does not roll over'), then '$0.0003 / tool call' and '$0.003 / trigger event'; '10,000 requests per minute on Pro'; 'DPA self-serve in dashboard'. Enterprise 'Custom' ('SSO / SCIM / customer-managed keys', 'MSA / DPA / BAA included'). Add-ons on Pro: BAA '$0.0003 per tool call on Pro', ZDR '+$0.0001 / tool call' (the page lists no per-trigger ZDR charge), IP allowlist '$0.0001 per tool call on Pro', advanced white-labeling 'One-time $0.30 per connection on Pro'. Composio-managed apps: 'Up to 20K of your 100K free, then $0.0005 / tool call'. Instant Tools 'include 20K free tool calls a month, then bill at an extra $0.0002 per tool call on top of the base rate'. 'Your earlier pricing is grandfathered through December 31, 2026.' See plan limits (/api/estimate) for the encoded plans.","strengths":["Huge tool catalog + MCP","Managed OAuth","Fast agent integration"],"weaknesses":["Abstraction over underlying tools","Pricing can surprise at scale","Dependency for critical workflows"],"tags":["mcp","tools","oauth","agent"],"scores":{"capabilityFit":8.2,"pricePerformance":7.5,"latencyReliability":7.8,"developerDx":8.8,"privacyPosture":7.2,"lockInRisk":5.8,"maturity":7.2,"overall":7.62,"kind":"curated_estimate","confidence":"high","asOf":"2026-10-19","sources":["Composio pricing: https://composio.dev/pricing","Composio: https://composio.dev","Composio docs: https://docs.composio.dev"],"notes":"Depot pick for tool/MCP router with managed OAuth across 1500+ apps. Official rate card is now public — Hobby is real free for light agents; Pro is the agent-scale default. Not a compute fabric; pair with E2B/Browserbase when you need sandboxes/browsers."},"lastReviewed":"2026-10-19","status":"active"},{"id":"browser-use","slug":"browser-use","name":"Browser Use Cloud","website":"https://browser-use.com","tagline":"Pick when you want an agent-native browser loop, not just a hosted Chromium. Browserbase still wins pure…","description":"Pick when you want an agent-native browser loop, not just a hosted Chromium. Browserbase still wins pure session polish. Strengths: OSS agent loop + cloud; NL agent mode and raw CDP; Active agent-browser ecosystem.","categoryIds":["cat-sandboxes-browsers","cat-search-scrape"],"jtbdIds":["web-agent","cloud-browser","llm-ready-extract","browser-agent-loop"],"regions":["global","us"],"pricingNotes":"As of 2026-09-14 (browser-use.com/pricing): No subscription. Browsers $0.02/hr (1-min min); residential proxy $5/GB (on by default); direct/BYO proxy $0.20/GB. Agent API = model cost + 20% service fee (or 20% of published rates if BYOK) + browser/traffic. $5 min top-up; $15 free credits once. Concurrency 10→50→250→500→1000 by lifetime spend. Hosted model examples (fee included): Claude Sonnet 5 $2.40/$12; GPT-5.6 Luna $0.24/$1.44; BU 2.0 $0.60/$3.50 per 1M. OSS library remains free to self-host.","strengths":["OSS agent loop + cloud","NL agent mode and raw CDP","Active agent-browser ecosystem"],"weaknesses":["Autonomy increases policy risk","Cloud claims need workload validation","Younger than Browserbase"],"tags":["browser","agent","cdp","oss"],"scores":{"capabilityFit":8,"pricePerformance":7.8,"latencyReliability":7,"developerDx":8.4,"privacyPosture":6.5,"lockInRisk":4,"maturity":6.2,"overall":7.4,"kind":"curated_estimate","confidence":"high","asOf":"2026-09-14","sources":["Browser Use pricing: https://browser-use.com/pricing","Browser Use: https://browser-use.com"],"notes":"Pick when you want an agent-native browser loop, not just a hosted Chromium. Browserbase still wins pure session polish."},"lastReviewed":"2026-09-14","status":"active"},{"id":"anchor-browser","slug":"anchor-browser","name":"Anchor Browser","website":"https://anchorbrowser.io","tagline":"Strong when bot-detection and authenticated computer-use matter. A/B vs Browserbase/Kernel.","description":"Strong when bot-detection and authenticated computer-use matter. A/B vs Browserbase/Kernel. Strengths: Humanized/stealth browsers; Auth profile story; MCP/SDK integration.","categoryIds":["cat-sandboxes-browsers"],"jtbdIds":["web-agent","cloud-browser","computer-use-browser"],"regions":["global","us"],"pricingNotes":"As of 2026-09-14 (docs.anchorbrowser.io/pricing): Free $5 credits/mo (5 concurrent); Starter $50/mo (25 conc); Growth $2,000/mo (200 conc); Enterprise custom. Infra (no-task): $0.01/browser create + $0.05/browser-hour + Anchor proxy $8/GB (BYO/proxyless egress $0.20/GB). AI steps from 2026-08-21: Light $0.01 / Standard $0.05 / High-perf $0.10 per step (BYOK flat $0.01/step). Growth adds SOC2/ISO27001 + EU/Asia/AU regions.","strengths":["Humanized/stealth browsers","Auth profile story","MCP/SDK integration"],"weaknesses":["Newer category player","Pricing less public","Overlaps Browserbase/Steel"],"tags":["browser","computer-use","mcp","stealth"],"scores":{"capabilityFit":7.8,"pricePerformance":7.2,"latencyReliability":7.2,"developerDx":8,"privacyPosture":7,"lockInRisk":4.8,"maturity":5.8,"overall":7.17,"kind":"curated_estimate","confidence":"high","asOf":"2026-09-14","sources":["Anchor Browser pricing: https://docs.anchorbrowser.io/pricing","Anchor Browser: https://anchorbrowser.io","Anchor docs: https://docs.anchorbrowser.io"],"notes":"Strong when bot-detection and authenticated computer-use matter. A/B vs Browserbase/Kernel."},"lastReviewed":"2026-09-14","status":"active"},{"id":"kernel","slug":"kernel","name":"Kernel","website":"https://www.kernel.sh","tagline":"Interesting Browserbase alternative with transparent per-second rates. Trial against your session mix.","description":"Interesting Browserbase alternative with transparent per-second rates. Trial against your session mix. Strengths: Fine-grained per-second browser pricing; Pools + Playwright exec; Agent-oriented APIs.","categoryIds":["cat-sandboxes-browsers"],"jtbdIds":["cloud-browser","web-agent","computer-use-browser"],"regions":["global","us"],"pricingNotes":"As of 2026-09-14 (kernel.sh/docs): Developer Free+$5 credits (5 concurrent); Hobbyist $30+$10 credits (10 conc); Start-Up $200+$50 credits (150 conc); Enterprise custom. Usage: headless $0.0000166667/s (~$0.06/hr), headful $0.0001333336/s (~$0.48/hr), GPU headful $0.0008000016/s (~$2.88/hr), app invocation $0.0000666668/s. Standby/idle pool time is free; proxies not billed. HIPAA on Enterprise.","strengths":["Fine-grained per-second browser pricing","Pools + Playwright exec","Agent-oriented APIs"],"weaknesses":["Newer brand","Need to model concurrency costs","Competes with Browserbase"],"tags":["browser","playwright","mcp"],"scores":{"capabilityFit":7.6,"pricePerformance":8,"latencyReliability":7.2,"developerDx":8.2,"privacyPosture":7,"lockInRisk":4.5,"maturity":5.8,"overall":7.32,"kind":"curated_estimate","confidence":"high","asOf":"2026-09-14","sources":["Kernel pricing docs: https://www.kernel.sh/docs/info/pricing","Kernel: https://www.kernel.sh","Kernel docs: https://www.kernel.sh/docs"],"notes":"Interesting Browserbase alternative with transparent per-second rates. Trial against your session mix."},"lastReviewed":"2026-09-14","status":"active"},{"id":"northflank","slug":"northflank","name":"Northflank","website":"https://northflank.com","tagline":"Choose when the agent needs a whole app platform, not just a code jail. E2B for pure sandboxes; Railway for…","description":"Choose when the agent needs a whole app platform, not just a code jail. E2B for pure sandboxes; Railway for simple apps. Strengths: Sandbox + PaaS + GPU + BYOC; microVM isolation; Full app platform.","categoryIds":["cat-cpu-compute","cat-sandboxes-browsers","cat-gpu-cloud"],"jtbdIds":["agent-backend","app-hosting","agent-sandbox","serverless-gpu","platform-sandbox"],"regions":["global","us"],"pricingNotes":"As of 2026-09-14 (northflank.com/pricing live): CPU $0.01667/vCPU-hr; Memory $0.00833/GB-hr; Disk $0.15/GB-mo; egress $0.06/GB. GPUs: L4 $0.80, A100 40GB $1.42, A100 80GB $1.76, H100 80GB $2.74, RTX PRO 6000 $3.00/hr. Predefined plans e.g. nf-compute-50 $12/mo, nf-compute-100-2 $24/mo. Sandbox free tier (2 services/1 DB/2 crons); PAYG self-serve; Enterprise BYOC. Prorated to the second.","strengths":["Sandbox + PaaS + GPU + BYOC","microVM isolation","Full app platform"],"weaknesses":["Broader than pure sandbox (complexity)","Closed platform","Less mindshare than Railway/Modal"],"tags":["paas","sandbox","gpu","byoc"],"scores":{"capabilityFit":8,"pricePerformance":7.4,"latencyReliability":7.8,"developerDx":8,"privacyPosture":7.5,"lockInRisk":5.5,"maturity":7.2,"overall":7.49,"kind":"curated_estimate","confidence":"high","asOf":"2026-09-14","sources":["Northflank: https://northflank.com","Northflank pricing: https://northflank.com/pricing","Agent sandbox landscape: https://shipengtao.com/en/agent-sandbox/"],"notes":"Choose when the agent needs a whole app platform, not just a code jail. E2B for pure sandboxes; Railway for simple apps."},"lastReviewed":"2026-09-14","status":"active"},{"id":"mem0","slug":"mem0","name":"Mem0","website":"https://mem0.ai","tagline":"Default hosted agent memory bolt-on. Model retrieval volume carefully; self-host OSS if caps hurt.","description":"Default hosted agent memory bolt-on. Model retrieval volume carefully; self-host OSS if caps hurt. Strengths: Drop-in agent memory API; OSS + hosted; Fast prototype path.","categoryIds":["cat-vector-memory"],"jtbdIds":["rag-memory","agent-memory-kv","prototype-rag","hosted-agent-memory"],"regions":["global","us"],"pricingNotes":"As of 2026-10-24 (mem0.ai/pricing, fetched 2026-10-24): Hosted memory layer. Hobby 'Free' ('10,000 add requests/ month', '1,000 retrieval requests/ month', '1 project'); Starter '$19 / month' ('50,000 add requests /month', '5,000 retrieval requests / month'); Pro '$249 / month' ('500,000 add requests /month', '50,000 retrieval requests /month', 'Graph memory (entity linking)'); Enterprise 'Custom' ('Unlimited add requests', 'Unlimited retrieval requests'). No per-request overage rate is published. The earlier Growth plan no longer appears on the page. Meters adds AND retrievals — read-heavy agents hit retrieval caps first. See plan limits (/api/estimate) for the encoded plans. OSS Apache-2.0 core self-hostable.","strengths":["Drop-in agent memory API","OSS + hosted","Fast prototype path"],"weaknesses":["Retrieval metering surprises","Graph gated to higher tiers","Not a full vector DB"],"tags":["memory","agent","oss","rag"],"scores":{"capabilityFit":7.8,"pricePerformance":8,"latencyReliability":7.4,"developerDx":8.8,"privacyPosture":7,"lockInRisk":5,"maturity":7.2,"overall":7.59,"kind":"curated_estimate","confidence":"high","asOf":"2026-09-14","sources":["Mem0: https://mem0.ai","Mem0 pricing analysis 2026: https://memnode.dev/articles/mem0-pricing-free-tier-limits-2026","Memory providers 2026: https://www.developersdigest.tech/blog/best-ai-agent-memory-providers-2026"],"notes":"Default hosted agent memory bolt-on. Model retrieval volume carefully; self-host OSS if caps hurt."},"lastReviewed":"2026-09-14","status":"active"},{"id":"zep","slug":"zep","name":"Zep","website":"https://www.getzep.com","tagline":"Pick when agents need temporal/entity graph memory, not just vector snippets. Mem0 for simpler recall.","description":"Pick when agents need temporal/entity graph memory, not just vector snippets. Mem0 for simpler recall. Strengths: Temporal graph memory; Strong for relational recall; OSS Graphiti path.","categoryIds":["cat-vector-memory"],"jtbdIds":["rag-memory","agent-memory-kv","hybrid-search-rag","temporal-graph-memory","hosted-agent-memory","episodic-long-term-memory"],"regions":["global","us"],"pricingNotes":"As of 2026-10-24 (getzep.com/pricing, fetched 2026-10-24): Temporal knowledge-graph memory (Graphiti). 'Self-serve plans are credit based.' Free: 'Start free with 10,000 credits a month.' ('No rollover or auto-topup.'). Flex '$125 / month billed monthly' ('50,000 credits included. Then $25 / 10,000 credits.', '30-day rollover', '600 requests per minute'). Flex Plus '$375 / month billed monthly' ('200,000 credits included. Then $75 / 40,000 credits.', '60-day rollover', '1,000 requests per minute'). Enterprise custom ('negotiated rates'). Annual billing: 'Save 17%' (the annual price is not published in the page text). Credits: '1 credit per Episode up to 350 bytes; +1 credit per additional 350 bytes (or part).'; '0 credits for retrieval, storage, threads, users, and graph storage.' See plan limits (/api/estimate) for the encoded plans. Self-host Graphiti + Neo4j/FalkorDB/Kuzu/Neptune.","strengths":["Temporal graph memory","Strong for relational recall","OSS Graphiti path"],"weaknesses":["Credit model needs modeling","Heavier than Mem0 for simple recall","Graph ops complexity if self-host"],"tags":["memory","graph","temporal","oss"],"scores":{"capabilityFit":8.2,"pricePerformance":7,"latencyReliability":7.6,"developerDx":7.8,"privacyPosture":7.8,"lockInRisk":5.2,"maturity":7,"overall":7.44,"kind":"curated_estimate","confidence":"high","asOf":"2026-09-14","sources":["Zep: https://www.getzep.com","Zep / Graphiti: https://github.com/getzep/graphiti","Memory comparison 2026: https://devtoollab.com/blog/ai-agent-memory-platforms"],"notes":"Pick when agents need temporal/entity graph memory, not just vector snippets. Mem0 for simpler recall."},"lastReviewed":"2026-09-14","status":"active"},{"id":"letta","slug":"letta","name":"Letta Cloud","website":"https://www.letta.com","tagline":"Choose when the agent IS a long-lived stateful runtime. For bolt-on memory use Mem0/Zep.","description":"Choose when the agent IS a long-lived stateful runtime. For bolt-on memory use Mem0/Zep. Strengths: Agent-owned memory model; OSS + cloud; Identity/stateful agents.","categoryIds":["cat-vector-memory","cat-sandboxes-browsers"],"jtbdIds":["rag-memory","agent-memory-kv","agent-backend","stateful-agent-runtime","hosted-agent-memory"],"regions":["global","us"],"pricingNotes":"As of 2026-09-14: Stateful agent runtime (ex-MemGPT). Free tier BYOK + limited agents; Pro ~$20/mo; API plan ~$20/mo + ~$0.10/active agent/mo + tool-exec ~$0.00015/s (secondary/roundup figures as stored then). As of 2026-10-25 https://www.letta.com/pricing answers HTTP 200 to plain curl via redirects (301 to https://docs.letta.com/letta-code/pricing, then 308 to https://docs.letta.com/pricing), not bot-blocked; the docs page quotes 'Pro For personal use. $20 /month', 'API Plan For teams building on the Letta API. $20 /month Unlimited agents $0.10 / active agent / mo $0.00015 / sec tool execution' and 'Free plans are limited to 3 stateful agents.' Stored figures are not re-priced here; any price change goes through the reviewed --catalog step. OSS self-host Apache-2.0. Re-verify console before commits.","strengths":["Agent-owned memory model","OSS + cloud","Identity/stateful agents"],"weaknesses":["More runtime than pure memory API","Pricing mixes hosting + usage","Learning curve"],"tags":["memory","stateful-agent","oss","memgpt"],"scores":{"capabilityFit":8,"pricePerformance":7.8,"latencyReliability":7.2,"developerDx":8,"privacyPosture":7.2,"lockInRisk":5.5,"maturity":6.8,"overall":7.4,"kind":"curated_estimate","confidence":"medium","asOf":"2026-09-14","sources":["Letta: https://www.letta.com","Letta GitHub: https://github.com/letta-ai/letta","Memory providers 2026: https://www.developersdigest.tech/blog/best-ai-agent-memory-providers-2026"],"notes":"Choose when the agent IS a long-lived stateful runtime. For bolt-on memory use Mem0/Zep."},"lastReviewed":"2026-09-14","status":"active"},{"id":"lancedb","slug":"lancedb","name":"LanceDB","website":"https://lancedb.com","tagline":"Great embedded/object-store vector layer. Pair with Mem0/Zep only if you need higher-level memory APIs.","description":"Great embedded/object-store vector layer. Pair with Mem0/Zep only if you need higher-level memory APIs. Strengths: Embedded + cloud flexibility; Object-storage-centric cost profile; Strong OSS momentum.","categoryIds":["cat-vector-memory","cat-object-storage"],"jtbdIds":["self-host-vector","cheap-vector-scale","prototype-rag","embedded-vector"],"regions":["global","us"],"pricingNotes":"As of 2026-09-14: Embedded + cloud vector DB on Lance format; excellent local/OSS DX and object-storage-friendly indexing. Cloud usage-based; OSS free. Popular as RAG/memory backend (e.g. with Cognee); not a hosted 'agent memory' product itself.","strengths":["Embedded + cloud flexibility","Object-storage-centric cost profile","Strong OSS momentum"],"weaknesses":["Less turnkey agent memory than Mem0","Ops for production clustering","Younger managed story vs Pinecone"],"tags":["vector","embedded","oss","lance"],"scores":{"capabilityFit":7.8,"pricePerformance":8.8,"latencyReliability":7.2,"developerDx":8.2,"privacyPosture":8,"lockInRisk":3.5,"maturity":6.8,"overall":7.78,"kind":"curated_estimate","confidence":"medium","asOf":"2026-09-14","sources":["LanceDB: https://lancedb.com","LanceDB docs: https://lancedb.github.io/lancedb/","LanceDB GitHub: https://github.com/lancedb/lancedb"],"notes":"Great embedded/object-store vector layer. Pair with Mem0/Zep only if you need higher-level memory APIs."},"lastReviewed":"2026-09-14","status":"active"},{"id":"duckdb","slug":"duckdb","name":"DuckDB (vector/extensions)","website":"https://duckdb.org","tagline":"Use for local analytics + light vector experiments. Production agent memory → pgvector/Qdrant/Mem0.","description":"Use for local analytics + light vector experiments. Production agent memory → pgvector/Qdrant/Mem0. Strengths: Zero-ops local analytics; SQL familiarity; Free/OSS.","categoryIds":["cat-vector-memory","cat-cpu-compute"],"jtbdIds":["prototype-rag","cheap-vector-scale","serverless-db","embedded-vector","sql-warehouse-agent-context"],"regions":["global","us"],"pricingNotes":"As of 2026-09-14: DuckDB is free MIT/open-source (local/embedded). Vector similarity via VSS + community extensions — not a hosted vector SaaS. MotherDuck is the hosted DuckDB warehouse (Lite $0 with 10 GB + 10 Pulse hrs; Business $250/org/mo + usage). Treat DuckDB as local OLAP + light vector for agent analytics, not a Pinecone replacement.","strengths":["Zero-ops local analytics","SQL familiarity","Free/OSS"],"weaknesses":["Not a managed vector service","Extension ecosystem evolving","Not multi-tenant agent memory"],"tags":["analytics","oss","vector-extension","local"],"scores":{"capabilityFit":6,"pricePerformance":9.5,"latencyReliability":7.5,"developerDx":8.5,"privacyPosture":9.5,"lockInRisk":2,"maturity":8.8,"overall":8,"kind":"curated_estimate","confidence":"high","asOf":"2026-09-14","sources":["DuckDB: https://duckdb.org","DuckDB VSS: https://duckdb.org/docs/stable/extensions/vss","MotherDuck pricing: https://motherduck.com/product/pricing/"],"notes":"Use for local analytics + light vector experiments. Production agent memory → pgvector/Qdrant/Mem0."},"lastReviewed":"2026-09-14","status":"active"},{"id":"promptfoo","slug":"promptfoo","name":"Promptfoo","website":"https://www.promptfoo.dev","tagline":"Default OSS eval harness. Braintrust when you want hosted collaboration; Promptfoo when you want local/CI…","description":"Default OSS eval harness. Braintrust when you want hosted collaboration; Promptfoo when you want local/CI control. Strengths: OSS + CI-native; Red teaming built-in; Multi-provider comparisons.","categoryIds":["cat-observability"],"jtbdIds":["evals","llm-tracing","oss-evals","adversarial-red-team-evals"],"regions":["global","us"],"pricingNotes":"As of 2026-10-24 (promptfoo.dev/pricing, fetched 2026-10-24): Community 'Free Forever' ('Our open-source tool, perfect for individual developers and small teams.') with 'Red teaming (10k probes/month)' and 'Run locally or self-host on your own infrastructure'; 'A probe is a single request made to your target system during red team testing.' Enterprise 'Custom' ('Contact us for a personalized quote.'; 'Enterprise customers can purchase additional probes for larger scale testing.'; 'Managed cloud deployment'). On-Premise 'Custom' ('Deployment on your own infrastructure'). No list price is published, so Enterprise price stays null.","strengths":["OSS + CI-native","Red teaming built-in","Multi-provider comparisons"],"weaknesses":["Less polished product UI than Braintrust","Enterprise features gated","You bring judge costs"],"tags":["evals","red-team","oss","ci"],"scores":{"capabilityFit":8.2,"pricePerformance":9.2,"latencyReliability":7.8,"developerDx":8.8,"privacyPosture":9,"lockInRisk":2.5,"maturity":8,"overall":8.42,"kind":"curated_estimate","confidence":"high","asOf":"2026-10-19","sources":["Promptfoo pricing: https://www.promptfoo.dev/pricing","Promptfoo: https://www.promptfoo.dev","LLM evals comparison 2026: https://qaskills.sh/blog/llm-evals-comparison-openai-promptfoo-ragas"],"notes":"Default OSS eval harness. Braintrust when you want hosted collaboration; Promptfoo when you want local/CI control."},"lastReviewed":"2026-10-19","status":"active"},{"id":"ragas","slug":"ragas","name":"Ragas","website":"https://docs.ragas.io","tagline":"Default RAG metric library. Pair with Promptfoo/Braintrust for broader agent evals.","description":"Default RAG metric library. Pair with Promptfoo/Braintrust for broader agent evals. Strengths: RAG-specific metrics; Free/OSS; Python-native.","categoryIds":["cat-observability"],"jtbdIds":["evals","rag-memory","rag-evals","oss-evals"],"regions":["global","us"],"pricingNotes":"As of 2026-09-14: Free open-source RAG evaluation library; cost is judge-LLM tokens only (~$0.50–$2 per 100 queries with cheap judges). No hosted SaaS required. Deep LangChain/LlamaIndex integration.","strengths":["RAG-specific metrics","Free/OSS","Python-native"],"weaknesses":["Not a full agent-trace platform","You operate harness","Less red-team than Promptfoo"],"tags":["evals","rag","oss"],"scores":{"capabilityFit":8,"pricePerformance":9.5,"latencyReliability":7.5,"developerDx":8.2,"privacyPosture":9.2,"lockInRisk":2,"maturity":7.8,"overall":8.33,"kind":"curated_estimate","confidence":"high","asOf":"2026-09-14","sources":["Ragas docs: https://docs.ragas.io","Ragas GitHub: https://github.com/explodinggradients/ragas","Best LLM eval tools 2026: https://deploybase.ai/articles/best-llm-evaluation-tools"],"notes":"Default RAG metric library. Pair with Promptfoo/Braintrust for broader agent evals."},"lastReviewed":"2026-09-14","status":"active"},{"id":"humanloop","slug":"humanloop","name":"Humanloop (sunset / Anthropic acquihire)","website":"https://humanloop.com","tagline":"Dead as an independent SKU. If a tutorial says Humanloop, translate to Braintrust / Promptfoo / Anthropic…","description":"Dead as an independent SKU. If a tutorial says Humanloop, translate to Braintrust / Promptfoo / Anthropic Console evals. Included so Depot can warn agents away. Strengths: Was a strong prompt CMS + HITL eval product; Team now inside Anthropic Console lineage; Historical docs still useful as a pattern library.","categoryIds":["cat-observability"],"jtbdIds":[],"regions":["global","us"],"pricingNotes":"HISTORICAL: Anthropic acqui-hired the Humanloop team (announced 2025-08). Standalone Humanloop platform shut down; prompt/eval workflows moved into Anthropic Console Workbench/Evaluations. Do not start new projects on humanloop.com.","strengths":["Was a strong prompt CMS + HITL eval product","Team now inside Anthropic Console lineage","Historical docs still useful as a pattern library"],"weaknesses":["Standalone product shut down after 2025 acquihire","No independent SLA or price card","Lock-in if you still have data only in Humanloop"],"tags":["evals","prompts","sunset","historical","anthropic"],"scores":{"capabilityFit":2,"pricePerformance":2.5,"latencyReliability":1.5,"developerDx":4,"privacyPosture":5,"lockInRisk":8.5,"maturity":3,"overall":2.68,"kind":"curated_estimate","confidence":"high","asOf":"2026-09-14","sources":["Anthropic nabs Humanloop team (TechCrunch): https://techcrunch.com/2025/08/13/anthropic-nabs-humanloop-team-as-competition-for-enterprise-ai-talent-heats-up/","Humanloop: https://humanloop.com"],"notes":"Dead as an independent SKU. If a tutorial says Humanloop, translate to Braintrust / Promptfoo / Anthropic Console evals. Included so Depot can warn agents away."},"lastReviewed":"2026-09-14","status":"deprecated"},{"id":"maxim","slug":"maxim","name":"Maxim AI","website":"https://www.getmaxim.ai","tagline":"Pick Maxim when pre-release agent simulation is the pain and seats are few. Braintrust for general evals…","description":"Pick Maxim when pre-release agent simulation is the pain and seats are few. Braintrust for general evals; Galileo for production guardrail monitoring; Promptfoo for CI/OSS. Strengths: Published seat/log tiers including free Developer; Strong simulation + offline/online eval story; CI/CD SDKs (Python/TS/Java/Go).","categoryIds":["cat-observability"],"jtbdIds":["evals","llm-tracing","production-guardrail-evals"],"regions":["global","us"],"pricingNotes":"As of 2026-09-14 (official getmaxim.ai / webflow pricing): Developer Free (≤3 seats, 10k logs/mo, 3-day retention); Professional $29/seat/mo (100k logs, simulation + online evals, 7-day retention); Business $49/seat/mo (500k logs, 30-day retention, RBAC, scheduled runs); Enterprise custom (SSO, in-VPC, SOC2/HIPAA/GDPR, BAA).","strengths":["Published seat/log tiers including free Developer","Strong simulation + offline/online eval story","CI/CD SDKs (Python/TS/Java/Go)"],"weaknesses":["Seat-based cost scales poorly for large eng orgs","Retention short on lower tiers","Less mindshare than LangSmith/Braintrust"],"tags":["evals","simulation","agents"],"scores":{"capabilityFit":8,"pricePerformance":7.2,"latencyReliability":7.8,"developerDx":8,"privacyPosture":8.2,"lockInRisk":5.5,"maturity":6.8,"overall":7.48,"kind":"curated_estimate","confidence":"high","asOf":"2026-09-14","sources":["Maxim AI pricing: https://webflow.getmaxim.ai/pricing","Agent simulation & evaluation: https://www.getmaxim.ai/evals/products/agent-simulation-evaluation","Maxim docs overview: https://www.getmaxim.ai/docs/introduction/overview"],"notes":"Pick Maxim when pre-release agent simulation is the pain and seats are few. Braintrust for general evals; Galileo for production guardrail monitoring; Promptfoo for CI/OSS."},"lastReviewed":"2026-09-14","status":"active"},{"id":"galileo","slug":"galileo","name":"Galileo","website":"https://www.galileo.ai","tagline":"Pick for production hallucination measurement + guardrails. Braintrust/Promptfoo for lighter eval loops.","description":"Pick for production hallucination measurement + guardrails. Braintrust/Promptfoo for lighter eval loops. Strengths: Strong eval metrics research; Guardrails loop; Generous free traces.","categoryIds":["cat-observability"],"jtbdIds":["evals","llm-tracing","cost-monitoring","production-guardrail-evals"],"regions":["global","us"],"pricingNotes":"As of 2026-09-14 (galileo.ai/pricing live): Free **$0/mo** — 5,000 traces/mo, unlimited users/custom evals. Pro **$100/mo** (annual; scales with traces) — 50,000 traces/mo, standard RBAC, advanced analytics, Slack support. Enterprise contact — unlimited traces, VPC/on-prem, SSO, real-time guardrails, 24/7.","strengths":["Strong eval metrics research","Guardrails loop","Generous free traces"],"weaknesses":["Enterprise features sales-gated","Hard to size ROI without sales","Heavier than Promptfoo"],"tags":["evals","guardrails","observability"],"scores":{"capabilityFit":8.4,"pricePerformance":6.2,"latencyReliability":8,"developerDx":7.8,"privacyPosture":8,"lockInRisk":6,"maturity":7.8,"overall":7.45,"kind":"curated_estimate","confidence":"high","asOf":"2026-09-14","sources":["Galileo pricing: https://www.galileo.ai/pricing","Galileo: https://www.galileo.ai","Galileo review 2026: https://llmtools.cc/tool/galileo/"],"notes":"Pick for production hallucination measurement + guardrails. Braintrust/Promptfoo for lighter eval loops."},"lastReviewed":"2026-09-14","status":"active"},{"id":"deepgram","slug":"deepgram","name":"Deepgram","website":"https://deepgram.com","tagline":"Default STT and strong voice-agent platform. Standard Flux TTS / Voice Agent rates live since 2026-09-15…","description":"Default STT and strong voice-agent platform. Standard Flux TTS / Voice Agent rates live since 2026-09-15 (confirmed 2026-09-22). Budget Voice Agent at $0.075/min Standard. Pair ElevenLabs/Cartesia if you need premium TTS voices. Strengths: Best-in-class STT pricing/quality; Voice Agent API; Clear public rate card; EU residency options.","categoryIds":["cat-media-gen","cat-inference-apis"],"jtbdIds":["speech-to-text","voice-agents","speech-synthesis","production-stt","voice-agent-platform","realtime-tts"],"regions":["global","us"],"pricingNotes":"As of 2026-09-22 (deepgram.com/pricing live): Flux TTS / Voice Agent promo ended; **Standard is live**. Flux TTS PAYG **$0.0450/1k** chars (Growth $0.0405); Aura-2 $0.030/1k; Aura-1 $0.015. Voice Agent Standard **$0.075/min** PAYG ($0.068 Growth); Standard BYO TTS $0.065/$0.051; Custom BYO LLM+TTS $0.050/$0.041; Advanced $0.163/$0.146. STT still shows limited-time streaming promos (Flux EN $0.0065/min PAYG; Nova-3 mono stream $0.0048). Buy Flux TTS get matching credits up to $500 through **2026-12-31**. Free $200 credit then PAYG.","strengths":["Best-in-class STT pricing/quality","Voice Agent API","Clear public rate card","EU residency options"],"weaknesses":["TTS less 'celebrity voice' than ElevenLabs","Voice Agent Standard jumped post-Flux-TTS promo (2026-09-13+)","Concurrency tiers"],"tags":["stt","tts","voice-agent","realtime"],"scores":{"capabilityFit":8.8,"pricePerformance":8.8,"latencyReliability":8.6,"developerDx":8.6,"privacyPosture":8.2,"lockInRisk":4,"maturity":9,"overall":8.5,"kind":"curated_estimate","confidence":"high","asOf":"2026-10-19","sources":["Deepgram pricing: https://deepgram.com/pricing","Deepgram: https://deepgram.com","Deepgram docs: https://developers.deepgram.com"],"notes":"Default STT and strong voice-agent platform. Standard Flux TTS / Voice Agent rates live since 2026-09-15 (confirmed 2026-09-22). Budget Voice Agent at $0.075/min Standard. Pair ElevenLabs/Cartesia if you need premium TTS voices."},"lastReviewed":"2026-10-19","status":"active"},{"id":"assemblyai","slug":"assemblyai","name":"AssemblyAI","website":"https://www.assemblyai.com","tagline":"Great when you need STT + audio intelligence features. Deepgram usually wins raw $/min STT.","description":"Great when you need STT + audio intelligence features. Deepgram usually wins raw $/min STT. Strengths: Excellent speech understanding add-ons; Developer DX; Voice agent packaging.","categoryIds":["cat-media-gen","cat-inference-apis"],"jtbdIds":["speech-to-text","voice-agents","production-stt","voice-agent-platform"],"regions":["global","us"],"pricingNotes":"As of 2026-09-14: Async Universal tiers roughly $0.15–$0.21/hr; realtime higher (~$0.45/hr class); Voice Agent API ~$4.50/hr all-inclusive packaging on public pages. New accounts often get ~$50 credits. Strong LLM-ready transcript features (summaries, PII redaction).","strengths":["Excellent speech understanding add-ons","Developer DX","Voice agent packaging"],"weaknesses":["Often pricier than Deepgram raw STT","Voice agent bundle math varies","Less TTS depth"],"tags":["stt","voice-agent","transcription"],"scores":{"capabilityFit":8.6,"pricePerformance":7.2,"latencyReliability":8.4,"developerDx":8.8,"privacyPosture":8,"lockInRisk":4.2,"maturity":8.8,"overall":8.11,"kind":"curated_estimate","confidence":"high","asOf":"2026-09-14","sources":["AssemblyAI pricing: https://www.assemblyai.com/pricing","AssemblyAI docs billing: https://www.assemblyai.com/docs/billing-and-pricing","Voice agent API comparison: https://www.assemblyai.com/blog/best-voice-agent-api"],"notes":"Great when you need STT + audio intelligence features. Deepgram usually wins raw $/min STT."},"lastReviewed":"2026-09-14","status":"active"},{"id":"cartesia","slug":"cartesia","name":"Cartesia","website":"https://cartesia.ai","tagline":"Top pick for realtime low-latency voice agents. ElevenLabs for voice library breadth.","description":"Top pick for realtime low-latency voice agents. ElevenLabs for voice library breadth. Strengths: Ultra-low-latency TTS; Simple credit plans; Voice agents + STT in one.","categoryIds":["cat-media-gen"],"jtbdIds":["speech-synthesis","voice-agents","realtime-tts","voice-agent-platform"],"regions":["global","us"],"pricingNotes":"As of 2026-09-14: Credit plans — Free $0 (20k credits), Pro $5 (100k), Startup $49 (1.25M), Scale $299 (8M) + prepaid agent minutes. Sonic TTS + Ink STT + voice agents. Cartesia phone numbers ~$0.014/min. Ultra-low-latency voice narrative.","strengths":["Ultra-low-latency TTS","Simple credit plans","Voice agents + STT in one"],"weaknesses":["Credit math less intuitive than per-char","Younger than ElevenLabs","Voice catalog still growing"],"tags":["tts","realtime","voice-agent"],"scores":{"capabilityFit":8.4,"pricePerformance":8,"latencyReliability":7.8,"developerDx":8.4,"privacyPosture":7.5,"lockInRisk":4.5,"maturity":7,"overall":7.81,"kind":"curated_estimate","confidence":"high","asOf":"2026-09-14","sources":["Cartesia pricing: https://www.cartesia.ai/pricing","Cartesia: https://cartesia.ai","Cartesia docs: https://docs.cartesia.ai"],"notes":"Top pick for realtime low-latency voice agents. ElevenLabs for voice library breadth."},"lastReviewed":"2026-09-14","status":"active"},{"id":"playht","slug":"playht","name":"PlayHT / PlayAI (sunset / Meta)","website":"https://play.ht","tagline":"Dead independent SKU. New TTS → Cartesia/ElevenLabs/Deepgram. Included as a warning entry.","description":"Dead independent SKU. New TTS → Cartesia/ElevenLabs/Deepgram. Included as a warning entry. Strengths: Historically competitive TTS/cloning API; Now inside Meta voice stack; Warning catalog entry.","categoryIds":["cat-media-gen"],"jtbdIds":[],"regions":["global","us"],"pricingNotes":"HISTORICAL: PlayAI (ex-PlayHT) acquired by Meta ~July 2025; standalone subscriptions/API no longer active. Legacy figures (~$20/1M chars Play 3.0 Mini, ~$60/1M PlayDialog) are not usable. Do not build new TTS on play.ht / play.ai.","strengths":["Historically competitive TTS/cloning API","Now inside Meta voice stack","Warning catalog entry"],"weaknesses":["Standalone API gone","No public successor SKU for indie builders","Do not migrate toward PlayAI"],"tags":["tts","voice","sunset","historical","meta"],"scores":{"capabilityFit":2,"pricePerformance":2,"latencyReliability":1,"developerDx":3,"privacyPosture":4,"lockInRisk":9,"maturity":2.5,"overall":2.18,"kind":"curated_estimate","confidence":"high","asOf":"2026-09-14","sources":["PlayAI no longer active: https://www.therundown.ai/tools/playai","PlayHT: https://play.ht"],"notes":"Dead independent SKU. New TTS → Cartesia/ElevenLabs/Deepgram. Included as a warning entry."},"lastReviewed":"2026-09-14","status":"deprecated"},{"id":"temporal-cloud","slug":"temporal-cloud","name":"Temporal Cloud","website":"https://temporal.io","tagline":"Default durable orchestration for serious long-running agents. Inngest/Trigger.dev for lighter DX.","description":"Default durable orchestration for serious long-running agents. Inngest/Trigger.dev for lighter DX. Strengths: Battle-tested durable workflows; OSS + cloud; Excellent failure recovery.","categoryIds":["cat-cpu-compute"],"jtbdIds":["agent-backend","notification-orchestration","durable-agent-workflows","hitl-approval-workflows"],"regions":["global","us"],"pricingNotes":"As of 2026-10-21 (temporal.io/pricing and docs.temporal.io/cloud/pricing, fetched 2026-10-21): Durable execution for long-running agent workflows. Developer plan is Pay-As-You-Go with no base monthly fee ('No base fee - 10% of Usage Spend') and Actions at $50 per million at every volume. Business is 'Greater of $500/mo or 10% of usage', includes 2.5M Actions, 2.5 GB Active Storage and 100 GB Retained Storage, with progressive Action discounts ('Self service down to $25 per million actions'). Storage is metered separately: $0.042 per GBh active, $0.00105 per GBh retained. Enterprise and Mission Critical are priced annually via Sales. temporal.io/pricing also lists 'Free $150 credits for 90 days'. The earlier 'Cloud Essentials' plan no longer appears on either page. See plan limits (/api/estimate) for the encoded plans. OSS Temporal free to self-host. Best for reliable multi-step agent orchestration with retries/state.","strengths":["Battle-tested durable workflows","OSS + cloud","Excellent failure recovery"],"weaknesses":["Steeper learning curve","Overkill for simple cron agents","Cloud cost vs Inngest simplicity"],"tags":["orchestration","durable-execution","workflows"],"scores":{"capabilityFit":9,"pricePerformance":6.8,"latencyReliability":9.2,"developerDx":7.2,"privacyPosture":8.2,"lockInRisk":5,"maturity":9.2,"overall":8.02,"kind":"curated_estimate","confidence":"high","asOf":"2026-09-14","sources":["Temporal: https://temporal.io","Temporal Cloud pricing: https://docs.temporal.io/cloud/pricing","Temporal docs: https://docs.temporal.io"],"notes":"Default durable orchestration for serious long-running agents. Inngest/Trigger.dev for lighter DX."},"lastReviewed":"2026-09-14","status":"active"},{"id":"inngest","slug":"inngest","name":"Inngest","website":"https://www.inngest.com","tagline":"Best DX default for durable agent workflows in TypeScript/Python. Temporal when you need harder multi-lang…","description":"Best DX default for durable agent workflows in TypeScript/Python. Temporal when you need harder multi-lang scale. Strengths: Great DX for durable steps; Fair free tier; OSS escape hatch.","categoryIds":["cat-cpu-compute"],"jtbdIds":["agent-backend","notification-orchestration","event-driven-agent-jobs","durable-agent-workflows","serverless-agent-queue"],"regions":["global","us"],"pricingNotes":"As of 2026-09-14: Hobby $0 (50k executions); Pro from $99/mo (1M executions, PAYG to 20M); Enterprise custom. OSS self-host available. Step functions with retries, fan-out, and observability aimed at event-driven agents.","strengths":["Great DX for durable steps","Fair free tier","OSS escape hatch"],"weaknesses":["Less multi-language maturity than Temporal","Execution pricing to model","Younger at extreme scale"],"tags":["orchestration","durable-execution","events"],"scores":{"capabilityFit":8.2,"pricePerformance":8,"latencyReliability":8.4,"developerDx":9,"privacyPosture":7.8,"lockInRisk":4.2,"maturity":7.8,"overall":8.07,"kind":"curated_estimate","confidence":"high","asOf":"2026-09-14","sources":["Inngest pricing: https://www.inngest.com/pricing","Inngest: https://www.inngest.com","Inngest docs: https://www.inngest.com/docs"],"notes":"Best DX default for durable agent workflows in TypeScript/Python. Temporal when you need harder multi-lang scale."},"lastReviewed":"2026-09-14","status":"active"},{"id":"trigger-dev","slug":"trigger-dev","name":"Trigger.dev","website":"https://trigger.dev","tagline":"Pick for TS app background agent jobs. Inngest for event meshes; Temporal for polyglot durability.","description":"Pick for TS app background agent jobs. Inngest for event meshes; Temporal for polyglot durability. Strengths: Superb TypeScript DX; Long-running tasks without timeouts; Fast onboarding.","categoryIds":["cat-cpu-compute"],"jtbdIds":["agent-backend","notification-orchestration","event-driven-agent-jobs","durable-agent-workflows","serverless-agent-queue"],"regions":["global","us"],"pricingNotes":"As of 2026-09-14 (trigger.dev/pricing): Free $0 with $5 credits, 20 concurrent; Hobby $10/mo; Pro $50/mo ($50 credits, 200+ concurrent then $10/mo per 50). Compute: Micro 0.25 vCPU $0.0000169/s; Small-1x $0.0000338/s; up to Large-2x 8 vCPU $0.0006800/s. Invocation $0.000025/prod run. DEV env free. Apache-2.0 self-host. HIPAA BAA add-on. Python via build extension inside TS tasks.","strengths":["Superb TypeScript DX","Long-running tasks without timeouts","Fast onboarding"],"weaknesses":["TS-centric","Smaller than Temporal/Inngest ecosystems","Pricing model to watch at scale"],"tags":["jobs","typescript","orchestration"],"scores":{"capabilityFit":7.8,"pricePerformance":8.2,"latencyReliability":8,"developerDx":9.2,"privacyPosture":7.5,"lockInRisk":4.8,"maturity":7.2,"overall":7.84,"kind":"curated_estimate","confidence":"high","asOf":"2026-09-14","sources":["Trigger.dev pricing: https://trigger.dev/pricing","Trigger.dev: https://trigger.dev","Trigger.dev docs: https://trigger.dev/docs"],"notes":"Pick for TS app background agent jobs. Inngest for event meshes; Temporal for polyglot durability."},"lastReviewed":"2026-09-14","status":"active"},{"id":"langgraph-cloud","slug":"langgraph-cloud","name":"LangGraph Platform / LangSmith Deploy","website":"https://www.langchain.com/langgraph","tagline":"Deploy here if your graphs already are LangGraph. Otherwise Inngest/Temporal + Langfuse to avoid stack…","description":"Deploy here if your graphs already are LangGraph. Otherwise Inngest/Temporal + Langfuse to avoid stack lock-in. Strengths: Native LangGraph deploy; Tied to LangSmith traces/evals; Good for graph agents.","categoryIds":["cat-cpu-compute","cat-observability"],"jtbdIds":["agent-backend","llm-tracing","evals","langgraph-deploy","durable-agent-workflows"],"regions":["global","us"],"pricingNotes":"As of 2026-09-14: LangGraph OSS free; hosted deployment is LangSmith Deployment. LangSmith Developer free; Plus ~$39/seat/mo with included traces + small serverless deployment, then LCU/LSU usage. Natural if already on LangChain/LangGraph.","strengths":["Native LangGraph deploy","Tied to LangSmith traces/evals","Good for graph agents"],"weaknesses":["LangChain ecosystem lock-in","Seat + usage pricing","Overkill if not using LangGraph"],"tags":["langgraph","orchestration","agents","langsmith"],"scores":{"capabilityFit":8.4,"pricePerformance":6.8,"latencyReliability":8,"developerDx":8.2,"privacyPosture":7.2,"lockInRisk":7.2,"maturity":8.2,"overall":7.49,"kind":"curated_estimate","confidence":"high","asOf":"2026-09-14","sources":["LangChain pricing: https://www.langchain.com/pricing","LangGraph: https://www.langchain.com/langgraph","LangGraph pricing breakdown 2026: https://www.truefoundry.com/blog/langgraph-pricing"],"notes":"Deploy here if your graphs already are LangGraph. Otherwise Inngest/Temporal + Langfuse to avoid stack lock-in."},"lastReviewed":"2026-09-14","status":"active"},{"id":"infisical","slug":"infisical","name":"Infisical","website":"https://infisical.com","tagline":"Depot default for agent secrets when you want OSS exit + Agent Proxy (placeholders instead of raw keys in the…","description":"Depot default for agent secrets when you want OSS exit + Agent Proxy (placeholders instead of raw keys in the agent). Doppler if you want seat-priced DX and free non-human identities. Strengths: MIT-licensed self-host + strong agent/machine identity story; Agent Proxy can keep real secrets off the agent process; US/EU cloud regions.","categoryIds":["cat-auth-payments"],"jtbdIds":["agent-secrets"],"regions":["us","eu","global"],"pricingNotes":"As of 2026-09-14: Secrets billed per identity (human+machine). Free: 5 identities + MIT self-host core. Pro ~$20/identity/mo annual ($23 monthly); Advanced ~$40/$46 with dynamic secrets + Agent Proxy for dynamic secrets; Enterprise custom. Agent Proxy for static secrets on all plans including Free.","strengths":["MIT-licensed self-host + strong agent/machine identity story","Agent Proxy can keep real secrets off the agent process","US/EU cloud regions"],"weaknesses":["Per-identity pricing hurts high machine-identity counts","Dynamic secrets gated to Advanced+","Younger ecosystem than Vault/1Password in enterprises"],"tags":["secrets","agent-identity","self-host","oss"],"scores":{"capabilityFit":8.8,"pricePerformance":8.2,"latencyReliability":8.2,"developerDx":8.8,"privacyPosture":9,"lockInRisk":3.5,"maturity":7.8,"overall":8.36,"kind":"curated_estimate","confidence":"high","asOf":"2026-09-14","sources":["Infisical pricing: https://infisical.com/pricing","Infisical: https://infisical.com"],"notes":"Depot default for agent secrets when you want OSS exit + Agent Proxy (placeholders instead of raw keys in the agent). Doppler if you want seat-priced DX and free non-human identities."},"lastReviewed":"2026-09-14","status":"active"},{"id":"doppler","slug":"doppler","name":"Doppler","website":"https://www.doppler.com","tagline":"Best secrets DX for most startups. Prefer Infisical when you need self-host or Agent Proxy that never hands…","description":"Best secrets DX for most startups. Prefer Infisical when you need self-host or Agent Proxy that never hands raw credentials to the agent runtime. Strengths: Excellent DX (CLI, syncs, service tokens); Agents/non-humans not billed as seats; Official MCP server for agent tooling.","categoryIds":["cat-auth-payments"],"jtbdIds":["agent-secrets","secret-rotation-automation"],"regions":["global"],"pricingNotes":"As of 2026-09-14: Per-seat — Developer free for 3 users then $8/user; Team $21/user/mo (14-day trial); Enterprise custom. Marketing: AI agents and non-human identities ride free (no agent seat fees). Official MCP server listed on Team+.","strengths":["Excellent DX (CLI, syncs, service tokens)","Agents/non-humans not billed as seats","Official MCP server for agent tooling"],"weaknesses":["Closed SaaS core (vs Infisical MIT self-host)","doppler run still injects plaintext into process env","Seat cost for large human teams"],"tags":["secrets","mcp","dx","service-tokens"],"scores":{"capabilityFit":8.6,"pricePerformance":8.4,"latencyReliability":8.8,"developerDx":9.4,"privacyPosture":8.5,"lockInRisk":5,"maturity":8.8,"overall":8.47,"kind":"curated_estimate","confidence":"high","asOf":"2026-09-14","sources":["Doppler pricing: https://www.doppler.com/pricing","Doppler: https://www.doppler.com"],"notes":"Best secrets DX for most startups. Prefer Infisical when you need self-host or Agent Proxy that never hands raw credentials to the agent runtime."},"lastReviewed":"2026-09-14","status":"active"},{"id":"onepassword-secrets","slug":"onepassword-secrets","name":"1Password Secrets Automation","website":"https://developer.1password.com/docs/service-accounts","tagline":"Use when the company already lives in 1Password Business. Don't buy 1Password just for agents — pick…","description":"Use when the company already lives in 1Password Business. Don't buy 1Password just for agents — pick Doppler/Infisical instead. Strengths: Zero-knowledge brand trust already in many orgs; Service Accounts for CI/agents without human login; Bundled with existing Business seats.","categoryIds":["cat-auth-payments"],"jtbdIds":["agent-secrets"],"regions":["global"],"pricingNotes":"As of 2026-09-14: Secrets Automation / Service Accounts included with 1Password Business (~$8.99/user/mo annual class) and Enterprise — no separate secrets SKU. Connect servers + Service Accounts for programmatic vault access.","strengths":["Zero-knowledge brand trust already in many orgs","Service Accounts for CI/agents without human login","Bundled with existing Business seats"],"weaknesses":["Weaker agent-native story than Infisical/Doppler","Tied to 1Password Business/Enterprise packaging","Less popular for pure machine-identity fleets"],"tags":["secrets","service-accounts","zero-knowledge"],"scores":{"capabilityFit":7.8,"pricePerformance":7.5,"latencyReliability":9,"developerDx":7.8,"privacyPosture":9.2,"lockInRisk":5.5,"maturity":9.2,"overall":7.98,"kind":"curated_estimate","confidence":"high","asOf":"2026-09-14","sources":["1Password Secrets Automation: https://www.1password.dev/secrets-automation","1Password Business pricing: https://1password.com/pricing/business"],"notes":"Use when the company already lives in 1Password Business. Don't buy 1Password just for agents — pick Doppler/Infisical instead."},"lastReviewed":"2026-09-14","status":"active"},{"id":"aws-secrets-manager","slug":"aws-secrets-manager","name":"AWS Secrets Manager","website":"https://aws.amazon.com/secrets-manager/","tagline":"Default only if the agent already runs inside AWS with IAM roles. Otherwise Doppler/Infisical.","description":"Default only if the agent already runs inside AWS with IAM roles. Otherwise Doppler/Infisical. Strengths: Native IAM + VPC endpoint story; Automatic rotation patterns; Global regional availability.","categoryIds":["cat-auth-payments"],"jtbdIds":["agent-secrets"],"regions":["global","us","eu"],"pricingNotes":"As of 2026-09-14: ~$0.40/secret/month + $0.05 per 10k API calls (classic AWS list); rotation via Lambda. Fine when already on AWS IAM.","strengths":["Native IAM + VPC endpoint story","Automatic rotation patterns","Global regional availability"],"weaknesses":["Awkward multi-cloud / local-dev DX","Per-secret fees encourage secret sprawl pain","Not agent-first (no MCP/Agent Proxy polish)"],"tags":["secrets","aws","iam","rotation"],"scores":{"capabilityFit":8,"pricePerformance":7,"latencyReliability":9.4,"developerDx":6.5,"privacyPosture":8.8,"lockInRisk":7,"maturity":9.5,"overall":7.69,"kind":"curated_estimate","confidence":"high","asOf":"2026-09-14","sources":["AWS Secrets Manager: https://aws.amazon.com/secrets-manager/","AWS Secrets Manager pricing: https://aws.amazon.com/secrets-manager/pricing/"],"notes":"Default only if the agent already runs inside AWS with IAM roles. Otherwise Doppler/Infisical."},"lastReviewed":"2026-09-14","status":"active"},{"id":"ovhcloud-ai","slug":"ovhcloud-ai","name":"OVHcloud AI Endpoints","website":"https://endpoints.ai.cloud.ovh.net","tagline":"Strong EU budget open-model alternative. Mistral remains the EU product default; OVH when price + French…","description":"Strong EU budget open-model alternative. Mistral remains the EU product default; OVH when price + French cloud residency matter more than brand DX. Strengths: French/EU sovereign cloud posture; OpenAI-compatible; includes guard + ASR models; Aggressive open-model token pricing vs US hosts.","categoryIds":["cat-inference-apis"],"jtbdIds":["eu-sovereign-llm","open-model-inference","content-moderation","speech-to-text"],"regions":["eu"],"pricingNotes":"As of 2026-09-14: Official catalog endpoints.ai.cloud.ovh.net (EU OpenAI-compat). CloudPrice snapshot of OVH list: Llama 3.3 70B ~$0.67/$0.67 per 1M; GPT-OSS-120B $0.08/$0.40; GPT-OSS-20B $0.04/$0.15; Llama 3.1 8B $0.10/$0.10; Mistral Small 3.2 24B $0.09/$0.28; Qwen3 32B $0.08/$0.23. €→USD conversion moves; re-export live € table from console before large commits.","strengths":["French/EU sovereign cloud posture","OpenAI-compatible; includes guard + ASR models","Aggressive open-model token pricing vs US hosts"],"weaknesses":["Catalog/UX less polished than Mistral La Plateforme","US latency and ecosystem weaker","Model freshness can lag US neoclouds"],"tags":["eu","sovereign","open-models","moderation"],"scores":{"capabilityFit":7.2,"pricePerformance":8.8,"latencyReliability":7.8,"developerDx":7.2,"privacyPosture":9.2,"lockInRisk":4,"maturity":7.5,"overall":7.72,"kind":"curated_estimate","confidence":"high","asOf":"2026-09-14","sources":["OVHcloud AI Endpoints catalog: https://www.ovhcloud.com/en/public-cloud/ai-endpoints/catalog/","OVHcloud AI Endpoints docs: https://endpoints.ai.cloud.ovh.net/docs","OVHcloud model prices (CloudPrice): https://cloudprice.net/models/providers/ovhcloud"],"notes":"Strong EU budget open-model alternative. Mistral remains the EU product default; OVH when price + French cloud residency matter more than brand DX."},"lastReviewed":"2026-09-14","status":"active"},{"id":"scaleway-generative","slug":"scaleway-generative","name":"Scaleway Generative APIs","website":"https://www.scaleway.com/en/generative-apis/","tagline":"Best-documented EU open-model serverless API after Mistral. Use when you want Paris residency + published €…","description":"Best-documented EU open-model serverless API after Mistral. Use when you want Paris residency + published € rates without Mistral commercial models. Strengths: Clear public € token table + EU (Paris) residency; OpenAI-compatible; function calling + structured outputs; Privacy: no train/reuse of prompts/outputs claimed.","categoryIds":["cat-inference-apis"],"jtbdIds":["eu-sovereign-llm","open-model-inference","embeddings","speech-to-text"],"regions":["eu"],"pricingNotes":"As of 2026-09-14 (official pricing table): Pay-per-1M tokens in Paris; 1M free tokens for new customers; examples — llama-3.3-70b €0.90/€0.90, gpt-oss-120b €0.15/€0.60, qwen3-coder-30b €0.20/€0.80, whisper-large-v3 €0.003/audio minute, embeddings from €0.10/M. Batch discounts for non-realtime. Dedicated Managed Inference alternative for reserved throughput.","strengths":["Clear public € token table + EU (Paris) residency","OpenAI-compatible; function calling + structured outputs","Privacy: no train/reuse of prompts/outputs claimed"],"weaknesses":["Single-region (Paris) today","Frontier closed models not the pitch","Quotas require KYC/credit card for full limits"],"tags":["eu","sovereign","open-models","paris"],"scores":{"capabilityFit":7.8,"pricePerformance":8.6,"latencyReliability":8,"developerDx":8.2,"privacyPosture":9.3,"lockInRisk":3.8,"maturity":8,"overall":8.09,"kind":"curated_estimate","confidence":"high","asOf":"2026-09-14","sources":["Scaleway Generative APIs: https://www.scaleway.com/en/generative-apis/","Scaleway MaaS pricing: https://www.scaleway.com/en/pricing/model-as-a-service/"],"notes":"Best-documented EU open-model serverless API after Mistral. Use when you want Paris residency + published € rates without Mistral commercial models."},"lastReviewed":"2026-09-14","status":"active"},{"id":"stackit-ai","slug":"stackit-ai","name":"STACKIT AI Model Serving","website":"https://stackit.com/en/products/data-ai/stackit-ai-model-serving","tagline":"Yes, STACKIT is real. Choose for German data-sovereignty RFPs. Not an indie default — onboarding friction is…","description":"Yes, STACKIT is real. Choose for German data-sovereignty RFPs. Not an indie default — onboarding friction is real. Strengths: Confirmed German sovereign cloud LLM API; Published € token SKUs; OpenAI-compat + RAG template.","categoryIds":["cat-inference-apis"],"jtbdIds":["eu-sovereign-llm","open-model-inference","embeddings"],"regions":["eu"],"pricingNotes":"As of 2026-09-14: Real Schwarz Group / STACKIT product (Germany South EU01). Token tiers — Standard €0.15 in / €0.25 out per 1M; Plus €0.45 / €0.65; Premium €1.50 / €1.75; embeddings €0.02–€0.08/M. OpenAI-compatible endpoint. Free 30-day trial after manual approval (~24h). Data not stored/used for training per product page.","strengths":["Confirmed German sovereign cloud LLM API","Published € token SKUs","OpenAI-compat + RAG template"],"weaknesses":["Manual trial gating / slower indie onboarding","Tier-based model packaging vs named model rates","Smaller ecosystem than Mistral/Scaleway"],"tags":["eu","sovereign","germany","schwarz"],"scores":{"capabilityFit":7,"pricePerformance":8,"latencyReliability":7.8,"developerDx":6.5,"privacyPosture":9.5,"lockInRisk":4.5,"maturity":7.2,"overall":7.39,"kind":"curated_estimate","confidence":"high","asOf":"2026-09-14","sources":["STACKIT AI Model Serving: https://stackit.com/en/products/data-ai/stackit-ai-model-serving","Available shared models docs: https://docs.stackit.cloud/products/data-and-ai/ai-model-serving/basics/available-shared-models/"],"notes":"Yes, STACKIT is real. Choose for German data-sovereignty RFPs. Not an indie default — onboarding friction is real."},"lastReviewed":"2026-09-14","status":"active"},{"id":"openai-moderation","slug":"openai-moderation","name":"OpenAI Moderation","website":"https://platform.openai.com/docs/guides/moderation","tagline":"Default first-pass content filter for agent I/O. Pair with Llama Guard (Groq) or NeMo Guardrails when you…","description":"Default first-pass content filter for agent I/O. Pair with Llama Guard (Groq) or NeMo Guardrails when you need open/custom policies. Strengths: Free, fast, zero infra; Text + image multimodal classification; Inline with generation APIs.","categoryIds":["cat-inference-apis","cat-observability"],"jtbdIds":["content-moderation","llm-input-guardrails","prompt-injection-firewall"],"regions":["global","us"],"pricingNotes":"As of 2026-09-14: Free moderation endpoint — omni-moderation-latest for text+images (images ≤20 MB). Also inline moderation on Responses/Chat Completions via moderation.model. Categories include hate, harassment, self-harm, sexual, violence, illicit, etc. Not customizable / not self-hostable.","strengths":["Free, fast, zero infra","Text + image multimodal classification","Inline with generation APIs"],"weaknesses":["Fixed OpenAI taxonomy — not your policy engine","Cannot self-host or fine-tune categories","Audio not classified"],"tags":["moderation","safety","free","policy"],"scores":{"capabilityFit":8.2,"pricePerformance":9.8,"latencyReliability":9.2,"developerDx":9.4,"privacyPosture":7.5,"lockInRisk":5.5,"maturity":9.5,"overall":8.62,"kind":"curated_estimate","confidence":"high","asOf":"2026-09-14","sources":["OpenAI Moderation guide: https://developers.openai.com/api/docs/guides/moderation","Moderation API: https://platform.openai.com/docs/guides/moderation"],"notes":"Default first-pass content filter for agent I/O. Pair with Llama Guard (Groq) or NeMo Guardrails when you need open/custom policies."},"lastReviewed":"2026-09-14","status":"active"},{"id":"nvidia-nemo-guardrails","slug":"nvidia-nemo-guardrails","name":"NVIDIA NeMo Guardrails","website":"https://docs.nvidia.com/nemo/guardrails/","tagline":"When OpenAI Moderation taxonomy is not enough and you need programmable rails. Host Llama Guard on Groq for…","description":"When OpenAI Moderation taxonomy is not enough and you need programmable rails. Host Llama Guard on Groq for the cheap classifier leaf. Strengths: Programmable Colang/rails policy engine; Composes with Llama Guard / NIM models; Serious production guardrail toolkit.","categoryIds":["cat-observability","cat-inference-apis"],"jtbdIds":["llm-input-guardrails","content-moderation","production-guardrail-evals","prompt-injection-firewall"],"regions":["global","us"],"pricingNotes":"As of 2026-09-14: Open-source NeMo Guardrails library (self-host) + NeMo microservices / NIM integration paths. Cost = your GPUs or NVIDIA API Catalog / NIM inference rates when routing guard models. Supports Llama Guard and other engines via OpenAI-compatible base_url.","strengths":["Programmable Colang/rails policy engine","Composes with Llama Guard / NIM models","Serious production guardrail toolkit"],"weaknesses":["Ops-heavy vs free OpenAI Moderation","DX steeper than drop-in classifiers","Hosted packaging split across NIM/microservices"],"tags":["guardrails","safety","nvidia","oss"],"scores":{"capabilityFit":8.8,"pricePerformance":7.5,"latencyReliability":8,"developerDx":7,"privacyPosture":8.8,"lockInRisk":4.5,"maturity":8,"overall":7.87,"kind":"curated_estimate","confidence":"high","asOf":"2026-09-14","sources":["NeMo Guardrails docs: https://docs.nvidia.com/nemo/guardrails/","Llama Guard integration: https://docs.nvidia.com/nemo/guardrails/configure-guardrails/guardrail-catalog/third-party/llama-guard","NeMo microservices guardrails inference: https://docs.nvidia.com/nemo/microservices/latest/guardrails/concepts/inference.html"],"notes":"When OpenAI Moderation taxonomy is not enough and you need programmable rails. Host Llama Guard on Groq for the cheap classifier leaf."},"lastReviewed":"2026-09-14","status":"active"},{"id":"orb","slug":"orb","name":"Orb","website":"https://www.withorb.com","tagline":"Pick Orb when pricing complexity is the product (credits, tiers, simulations). Stripe meters first; Lago if…","description":"Pick Orb when pricing complexity is the product (credits, tiers, simulations). Stripe meters first; Lago if you want OSS. Strengths: Best-in-class pricing experiments / backtests for AI meters; Handles messy token+seat+tool hybrid plans; Mature fintech-grade metering.","categoryIds":["cat-auth-payments"],"jtbdIds":["usage-metering","billing"],"regions":["global","us"],"pricingNotes":"As of 2026-09-14: withorb.com/pricing remains sales-led — no published starter SKU or per-event list price. Quotes scale with billings/event volume. Strength is pricing simulation, backtesting, and multidimensional AI rate cards. Often paired with Stripe for collection. Round 4: still no anonymized public band.","strengths":["Best-in-class pricing experiments / backtests for AI meters","Handles messy token+seat+tool hybrid plans","Mature fintech-grade metering"],"weaknesses":["Opaque/sales-gated pricing","Overkill for simple Stripe Billing meters","Vendor lock on rating logic"],"tags":["billing","metering","usage-based","ai-saas"],"scores":{"capabilityFit":9.2,"pricePerformance":6.5,"latencyReliability":9,"developerDx":8.2,"privacyPosture":8.5,"lockInRisk":6.5,"maturity":8.8,"overall":8.03,"kind":"curated_estimate","confidence":"medium","asOf":"2026-09-14","sources":["Orb pricing: https://www.withorb.com/pricing","Orb: https://www.withorb.com","Lago vs Orb 2026: https://www.aibilling.dev/compare/lago-vs-orb"],"notes":"Pick Orb when pricing complexity is the product (credits, tiers, simulations). Stripe meters first; Lago if you want OSS."},"lastReviewed":"2026-09-14","status":"active"},{"id":"lago","slug":"lago","name":"Lago","website":"https://getlago.com","tagline":"Depot OSS pick for AI usage metering when you refuse to marry Stripe's meter model. Still collect cash with…","description":"Depot OSS pick for AI usage metering when you refuse to marry Stripe's meter model. Still collect cash with Stripe/Adyen. Strengths: OSS / self-host exit ramp; Purpose-built usage billing primitives; Vendor-neutral vs Stripe-only meters.","categoryIds":["cat-auth-payments"],"jtbdIds":["usage-metering","billing"],"regions":["global","eu"],"pricingNotes":"As of 2026-09-14: AGPLv3 OSS metering/billing — self-host free (infra+eng). Lago Cloud remains package/sales-led; docs FAQ does not publish a Cloud seat/event table. Third parties historically cite low-hundreds $/mo starters — unverified. Event-based usage + prepaid credits patterns common for AI APIs.","strengths":["OSS / self-host exit ramp","Purpose-built usage billing primitives","Vendor-neutral vs Stripe-only meters"],"weaknesses":["AGPL compliance diligence required","Cloud pricing less transparent than Stripe","You own more billing ops than Stripe Billing"],"tags":["billing","metering","oss","agpl"],"scores":{"capabilityFit":8.6,"pricePerformance":8.8,"latencyReliability":8,"developerDx":8,"privacyPosture":9,"lockInRisk":3,"maturity":7.8,"overall":8.3,"kind":"curated_estimate","confidence":"medium","asOf":"2026-09-14","sources":["Lago: https://getlago.com","Lago vs Orb: https://getlago.com/blog/lago-vs-orb","Lago docs pricing FAQ: https://docs.getlago.com/faq/pricing"],"notes":"Depot OSS pick for AI usage metering when you refuse to marry Stripe's meter model. Still collect cash with Stripe/Adyen."},"lastReviewed":"2026-09-14","status":"active"},{"id":"openmeter","slug":"openmeter","name":"OpenMeter (Kong Konnect Metering)","website":"https://openmeter.io","tagline":"Best when your agent product is an API gateway problem (entitlements + token meters). Orb for pricing…","description":"Best when your agent product is an API gateway problem (entitlements + token meters). Orb for pricing science; Lago for OSS; Stripe for simple SaaS. Strengths: Real-time entitlement gating for AI APIs; Native AI token metering examples; Backed by Kong distribution.","categoryIds":["cat-auth-payments"],"jtbdIds":["usage-metering","billing"],"regions":["global"],"pricingNotes":"As of 2026-09-14: OpenMeter Cloud repositioned as Kong Konnect Metering & Billing — real-time AI/API usage meters, entitlements, rate cards, invoicing. Strong fit for token metering via Kong plugin (meter_ai_token_usage). Pricing via Kong/Konnect packaging.","strengths":["Real-time entitlement gating for AI APIs","Native AI token metering examples","Backed by Kong distribution"],"weaknesses":["Product identity shifted into Kong Konnect","Less pure-billing polish than Orb","Pricing tied to Kong commercial motion"],"tags":["billing","metering","kong","entitlements"],"scores":{"capabilityFit":8.4,"pricePerformance":7.8,"latencyReliability":8.4,"developerDx":8.2,"privacyPosture":8.2,"lockInRisk":5.5,"maturity":8,"overall":7.93,"kind":"curated_estimate","confidence":"high","asOf":"2026-09-14","sources":["OpenMeter pricing / Konnect: https://openmeter.io/pricing","AI usage metering use case: https://openmeter.io/use-cases/ai","Kong meter AI tokens: https://developer.konghq.com/plugins/metering-and-billing/examples/meter-ai-tokens/"],"notes":"Best when your agent product is an API gateway problem (entitlements + token meters). Orb for pricing science; Lago for OSS; Stripe for simple SaaS."},"lastReviewed":"2026-09-14","status":"active"},{"id":"apollo","slug":"apollo","name":"Apollo.io","website":"https://www.apollo.io","tagline":"Default for sales/research agents that already think in Apollo credits. Prefer People Data Labs for…","description":"Default for sales/research agents that already think in Apollo credits. Prefer People Data Labs for high-volume developer pipelines without seat tax. Strengths: Huge B2B people/company graph for GTM agents; Enrichment + sequencing in one vendor; Mature API surface.","categoryIds":["cat-search-scrape"],"jtbdIds":["people-enrichment","agent-search"],"regions":["global","us"],"pricingNotes":"As of 2026-09-14: Sales platform + enrichment API. People enrichment typically 1 credit for email/demographics; mobile phone often +8 credits (up to ~9 credits/person when phone returned). Seat-based Apollo plans gate credit pools — check docs.apollo.io API pricing for current credit math.","strengths":["Huge B2B people/company graph for GTM agents","Enrichment + sequencing in one vendor","Mature API surface"],"weaknesses":["Credit/seat packaging opaque and expensive at phone scale","Sales-platform gravity (not pure data API)","Compliance/ToS limits on automated outreach"],"tags":["enrichment","b2b","people","gtm"],"scores":{"capabilityFit":8.8,"pricePerformance":6.8,"latencyReliability":8.4,"developerDx":8,"privacyPosture":6.2,"lockInRisk":5.5,"maturity":9,"overall":7.72,"kind":"curated_estimate","confidence":"high","asOf":"2026-09-14","sources":["Apollo API pricing: https://docs.apollo.io/docs/api-pricing","People Enrichment API: https://docs.apollo.io/reference/people-enrichment"],"notes":"Default for sales/research agents that already think in Apollo credits. Prefer People Data Labs for high-volume developer pipelines without seat tax."},"lastReviewed":"2026-09-14","status":"active"},{"id":"people-data-labs","slug":"people-data-labs","name":"People Data Labs","website":"https://www.peopledatalabs.com","tagline":"Depot pick for research-agent enrichment pipelines. Apollo when GTM workflow glue matters more than $/record.","description":"Depot pick for research-agent enrichment pipelines. Apollo when GTM workflow glue matters more than $/record. Strengths: API-first, usage-priced enrichment; Strong person + company datasets for agents; Better unit economics than seat-gated GTM suites.","categoryIds":["cat-search-scrape"],"jtbdIds":["people-enrichment"],"regions":["global","us"],"pricingNotes":"As of 2026-09-14: Developer-first person/company APIs; credit/record based (Person Enrichment often ~$0.22–$0.28 per successful match depending on tier/annual). No seat tax — better for automated research-agent pipelines. Verify current credit schedule in PDL help center.","strengths":["API-first, usage-priced enrichment","Strong person + company datasets for agents","Better unit economics than seat-gated GTM suites"],"weaknesses":["Raw data quality still needs validation layer","No built-in sequencing/CRM like Apollo","Privacy/compliance diligence required"],"tags":["enrichment","people","company","api-first"],"scores":{"capabilityFit":8.6,"pricePerformance":8.2,"latencyReliability":8.4,"developerDx":8.8,"privacyPosture":6.5,"lockInRisk":4,"maturity":8.6,"overall":8.14,"kind":"curated_estimate","confidence":"high","asOf":"2026-09-14","sources":["PDL pricing overview: https://support.peopledatalabs.com/hc/en-us/articles/23553812020891-Pricing-Overview","PDL pricing & credits: https://support.peopledatalabs.com/hc/en-us/articles/25794271805211-Pricing-credits","People Data Labs: https://www.peopledatalabs.com"],"notes":"Depot pick for research-agent enrichment pipelines. Apollo when GTM workflow glue matters more than $/record."},"lastReviewed":"2026-09-14","status":"active"},{"id":"openpipe","slug":"openpipe","name":"OpenPipe (training migrated)","website":"https://openpipe.ai","tagline":"Treat like Banana: historical warning entry. New fine-tunes → Together/Fireworks/Predibase/Modal.","description":"Treat like Banana: historical warning entry. New fine-tunes → Together/Fireworks/Predibase/Modal. Strengths: Pioneered simple request-log → SFT UX; ART open-source trainer still useful; Clear migration communication.","categoryIds":["cat-train-finetune"],"jtbdIds":[],"regions":["us"],"pricingNotes":"As of 2026-09-14: Hosted OpenPipe training/inference on openpipe.ai ended ~2026-07-30; workloads migrated toward W&B Inference. Open-source ART trainer remains. Do not start new hosted fine-tunes on OpenPipe cloud.","strengths":["Pioneered simple request-log → SFT UX","ART open-source trainer still useful","Clear migration communication"],"weaknesses":["Hosted training/inference retired","Migration to W&B changes vendor map","Stale tutorials still recommend OpenPipe cloud"],"tags":["finetune","sunset","wandb","historical"],"scores":{"capabilityFit":2.5,"pricePerformance":3,"latencyReliability":2,"developerDx":5,"privacyPosture":5,"lockInRisk":8,"maturity":4,"overall":3.26,"kind":"curated_estimate","confidence":"high","asOf":"2026-09-14","sources":["OpenPipe → W&B migration: https://openpipe.ai/blog/openpipe-training-inference-migrating-to-wandb","OpenPipe: https://openpipe.ai/"],"notes":"Treat like Banana: historical warning entry. New fine-tunes → Together/Fireworks/Predibase/Modal."},"lastReviewed":"2026-09-14","status":"deprecated"},{"id":"predibase","slug":"predibase","name":"Predibase","website":"https://predibase.com","tagline":"Alive. Prefer when you serve dozens of LoRA adapters on shared GPUs. Together/Fireworks remain simpler…","description":"Alive. Prefer when you serve dozens of LoRA adapters on shared GPUs. Together/Fireworks remain simpler defaults for one-off SFT. Strengths: LoRAX multi-adapter serving (many fine-tunes / GPU); Active fine-tune + RFT product vs OpenPipe sunset; Strong open-model post-train story.","categoryIds":["cat-train-finetune","cat-inference-apis"],"jtbdIds":["finetune","open-model-inference","lora-serving-distillation"],"regions":["us","global"],"pricingNotes":"As of 2026-09-14: Active managed LoRA/QLoRA + RL fine-tune (Rubrik-era Predibase); LoRAX OSS multi-adapter serving. Public token rates still not a clean first-party matrix — secondary sources cite serverless from ~$0.20/M and A10G-class GPUs ~$1.82/hr. Enterprise/VPC sales-led. Confirm console before quoting.","strengths":["LoRAX multi-adapter serving (many fine-tunes / GPU)","Active fine-tune + RFT product vs OpenPipe sunset","Strong open-model post-train story"],"weaknesses":["Pricing transparency uneven","Smaller mindshare than Together/Fireworks","Enterprise features sales-gated"],"tags":["finetune","lora","lorax","serving"],"scores":{"capabilityFit":8.4,"pricePerformance":8,"latencyReliability":7.8,"developerDx":8.2,"privacyPosture":7.8,"lockInRisk":4.5,"maturity":7.8,"overall":7.89,"kind":"curated_estimate","confidence":"medium","asOf":"2026-09-14","sources":["Predibase: https://predibase.com","LoRAX GitHub: https://github.com/predibase/lorax","Predibase plans (secondary): https://apis.io/plans/predibase/predibase-plans-pricing/"],"notes":"Alive. Prefer when you serve dozens of LoRA adapters on shared GPUs. Together/Fireworks remain simpler defaults for one-off SFT."},"lastReviewed":"2026-09-14","status":"active"},{"id":"scrapybara","slug":"scrapybara","name":"Scrapybara","website":"https://scrapybara.com","tagline":"Clear hosted desktop provider. E2B Desktop remains depot default if you already run E2B; Scrapybara when you…","description":"Clear hosted desktop provider. E2B Desktop remains depot default if you already run E2B; Scrapybara when you need Windows or turnkey desktop sessions. Strengths: Turnkey hosted desktops (incl. Windows) for computer-use; Session-oriented agent UX; Browser + OS control in one product.","categoryIds":["cat-sandboxes-browsers"],"jtbdIds":["desktop-computer-use","cloud-browser","web-agent"],"regions":["us","global"],"pricingNotes":"As of 2026-09-14: Official SDK 2.0 post publishes subscription SKUs — Basic $29/mo (100 compute hrs, 500 agent credits, 25 concurrent); Pro $99/mo (500 hrs, 2,500 credits, 100 concurrent); Enterprise self-host + 99.9% SLA. Meters: instance compute hours (Ubuntu/browser/Windows) + extra agent credits ~$0.04 (or BYOK to skip credits). Free tier historically ~10 compute hrs / 100 credits — confirm live dashboard. No separate public /pricing page; blog table is canonical.","strengths":["Turnkey hosted desktops (incl. Windows) for computer-use","Session-oriented agent UX","Browser + OS control in one product"],"weaknesses":["Younger / thinner production postmortems than E2B/Browserbase","Plan credit math can surprise","Less programmable than raw E2B Desktop SDK"],"tags":["desktop","computer-use","windows","sandbox"],"scores":{"capabilityFit":8.4,"pricePerformance":7.8,"latencyReliability":7.4,"developerDx":8.2,"privacyPosture":7.2,"lockInRisk":4.8,"maturity":6.5,"overall":7.58,"kind":"curated_estimate","confidence":"high","asOf":"2026-09-14","sources":["Scrapybara SDK 2.0 (plan table): https://scrapybara.com/blog/sdk2","Scrapybara: https://scrapybara.com","Scrapybara FinOps profile: https://apis.io/finops/scrapybara/scrapybara-finops/"],"notes":"Clear hosted desktop provider. E2B Desktop remains depot default if you already run E2B; Scrapybara when you need Windows or turnkey desktop sessions."},"lastReviewed":"2026-09-14","status":"active"},{"id":"smithery","slug":"smithery","name":"Smithery","website":"https://smithery.ai","tagline":"Public MCP server directory; hosted URLs, API, Connect, CLI and SDKs shut down 2026-10-31.","description":"Public (non-proxied) MCP server directory. Per https://smithery.ai/future (fetched 2026-10-24), its connection management services shut down on 2026-10-31: hosted server URLs (run.tools, server.smithery.ai), the Smithery API, Connect and Toolbox, accounts and API keys, and the CLI and SDKs stop working. No longer a Depot pick since 2026-10-24 (mcp-discovery -> mcp-registry, mcp-server-hosting -> cloudflare-workers); kept as a directory-only runner-up for mcp-discovery. Strengths: Public (non-proxied) MCP server directory; hosting ends 2026-10-31; Registry pages show direct MCP URLs for a curated set of verified servers; Faster path than raw registry metadata.","categoryIds":["cat-search-scrape","cat-sandboxes-browsers"],"jtbdIds":["mcp-discovery","agent-tool-router"],"regions":["global"],"pricingNotes":"As of 2026-10-24 (https://smithery.ai/future; banner on https://smithery.ai/pricing: 'Smithery's MCP Proxy and managed identity shut down on October 31.'): 'On October 31, 2026, we're shutting down Smithery's connection management services.' 'Smithery will remain as a public (non-proxied) MCP registry.' What stops working on October 31 includes 'Server URLs on run.tools and server.smithery.ai', 'Smithery API, Connect, and Toolbox' and 'Smithery accounts and API keys'. Paid plans: 'We're refunding every charge since August 5 and cancelling your subscription going forward.' 'You won't be charged again.' No current plan price is stated; the 2026-09-14 Pay as you Go and per-RPC prices were dropped because paid plans are being cancelled. The page states no price for browsing the public directory.","strengths":["Public (non-proxied) MCP server directory; hosting ends 2026-10-31","Registry pages show direct MCP URLs for a curated set of verified servers","Faster path than raw registry metadata"],"weaknesses":["Not the canonical registry of record","Server URLs, API, Connect, accounts, CLI and SDKs stop working 2026-10-31","Quality of listed servers varies"],"tags":["mcp","registry","tools","retirement-announced"],"scores":{"capabilityFit":8.2,"pricePerformance":8,"latencyReliability":7.6,"developerDx":8.8,"privacyPosture":7,"lockInRisk":5,"maturity":7,"overall":7.7,"kind":"curated_estimate","confidence":"high","asOf":"2026-09-14","sources":["Smithery pricing: https://smithery.ai/pricing","Smithery namespaces (plan limits): https://smithery.ai/docs/concepts/namespaces","Smithery: https://smithery.ai"],"notes":"Public (non-proxied) MCP server directory. Per https://smithery.ai/future (fetched 2026-10-24), its connection management services shut down on 2026-10-31: hosted server URLs (run.tools, server.smithery.ai), the Smithery API, Connect and Toolbox, accounts and API keys, and the CLI and SDKs stop working. No longer a Depot pick since 2026-10-24 (mcp-discovery -> mcp-registry, mcp-server-hosting -> cloudflare-workers); kept as a directory-only runner-up for mcp-discovery."},"lastReviewed":"2026-09-14","status":"active"},{"id":"mcp-registry","slug":"mcp-registry","name":"Official MCP Registry","website":"https://registry.modelcontextprotocol.io","tagline":"Source of truth for MCP server metadata and the Depot pick for mcp-discovery since 2026-10-24…","description":"Source of truth for MCP server metadata and the Depot pick for mcp-discovery since 2026-10-24 (replacing Smithery, whose connection services end 2026-10-31). Not a hosting runtime; see mcp-server-hosting. Strengths: Canonical metadata source of truth; Unauthenticated read-only REST API for agent discovery; Vendor-neutral.","categoryIds":["cat-search-scrape"],"jtbdIds":["mcp-discovery"],"regions":["global"],"pricingNotes":"As of 2026-10-24 (https://modelcontextprotocol.io/registry/registry-aggregators; https://github.com/modelcontextprotocol/registry/blob/main/docs/reference/api/official-registry-api.md): 'The MCP Registry provides an unauthenticated read-only REST API that aggregators can use to populate their data stores.' Base URL https://registry.modelcontextprotocol.io, GET /v0.1/servers ('List all servers.'), cursor pagination; 'Aggregators are expected to scrape data on a regular but infrequent basis (e.g., once per hour)'. Search: 'Case-insensitive substring search on server names' ('This is intentionally simple. For more advanced searching and filtering, use a subregistry.'). 'The MCP Registry is currently in preview. Breaking changes or data resets may occur before general availability.' 'The MCP Registry does not provide uptime or data durability guarantees'. Publishing: 'Publishing requires namespace-based authentication'. Neither page publishes a price or plan; no price is stated here. Not a hosting runtime: 'Aggregators are downstream consumers of the MCP Registry that provide additional value.' 'For example, a server marketplace that provides user ratings and security scanning.'","strengths":["Canonical metadata source of truth","Unauthenticated read-only REST API for agent discovery","Vendor-neutral"],"weaknesses":["Not a turnkey hosting/OAuth product","Sparse UX vs Glama/PulseMCP directories","Quality signals limited vs commercial indexes"],"tags":["mcp","registry","canonical"],"scores":{"capabilityFit":7,"pricePerformance":9.8,"latencyReliability":8.5,"developerDx":7.5,"privacyPosture":9,"lockInRisk":1,"maturity":7.5,"overall":8.19,"kind":"curated_estimate","confidence":"high","asOf":"2026-09-14","sources":["Official MCP Registry API reference: https://github.com/modelcontextprotocol/registry/blob/main/docs/reference/api/official-registry-api.md","Registry aggregators: https://modelcontextprotocol.io/registry/registry-aggregators","Registry base: https://registry.modelcontextprotocol.io"],"notes":"Source of truth for MCP server metadata and the Depot pick for mcp-discovery since 2026-10-24 (replacing Smithery, whose connection services end 2026-10-31). Not a hosting runtime; see mcp-server-hosting."},"lastReviewed":"2026-09-14","status":"active"},{"id":"aleph-alpha","slug":"aleph-alpha","name":"Aleph Alpha (PhariaAI)","website":"https://aleph-alpha.com","tagline":"Catalog for DACH/public-sector RFPs only. Depot EU default remains Mistral; OVH/Scaleway/STACKIT for…","description":"Catalog for DACH/public-sector RFPs only. Depot EU default remains Mistral; OVH/Scaleway/STACKIT for self-serve EU inference. Do not pick Aleph Alpha for a product agent that needs a credit-card API. Strengths: German sovereign stack (BSI/public-sector gravity, STACKIT partnership); On-prem / air-gapped PhariaOS for regulated buyers; PhariaInference has tools, embeddings, JSON, rerank.","categoryIds":["cat-inference-apis"],"jtbdIds":["eu-sovereign-llm","enterprise-llm"],"regions":["eu","de"],"pricingNotes":"As of 2026-09-14: Quote-only enterprise. No self-serve token page. Product is PhariaAI suite (PhariaAssistant/Studio/OS/Catch) + PhariaInference API (token minted in PhariaOS) — not a public Luminous API. Deployments: STACKIT SaaS, private cloud, on-prem; 4–12 week typical stand-up; 12–24 month contracts. Only published public-sector band is F13 ~€1.50/workplace/mo via Baden-Württemberg partner channel — not a commercial list price. Pharia-1 open weights on HF are research-only (Open Aleph License).","strengths":["German sovereign stack (BSI/public-sector gravity, STACKIT partnership)","On-prem / air-gapped PhariaOS for regulated buyers","PhariaInference has tools, embeddings, JSON, rerank"],"weaknesses":["No self-serve API or public rate card — unusable as an indie default","Weeks-long deployment vs Mistral/OVH minutes","Pivoted away from developer API-first in late 2024"],"tags":["eu","sovereign","germany","enterprise","on-prem"],"scores":{"capabilityFit":7.2,"pricePerformance":4.5,"latencyReliability":8,"developerDx":5.5,"privacyPosture":9.5,"lockInRisk":7,"maturity":7.8,"overall":6.58,"kind":"curated_estimate","confidence":"medium","asOf":"2026-09-14","sources":["Aleph Alpha: https://aleph-alpha.com","PhariaInference API docs: https://docs.aleph-alpha.com/phariaai-admin-guide/latest/pharia-inference/api/index.html","Aleph Alpha Pharia review 2026: https://www.teamazing.com/blog/aleph-alpha-pharia-review/"],"notes":"Catalog for DACH/public-sector RFPs only. Depot EU default remains Mistral; OVH/Scaleway/STACKIT for self-serve EU inference. Do not pick Aleph Alpha for a product agent that needs a credit-card API."},"lastReviewed":"2026-09-14","status":"active"},{"id":"lakera","slug":"lakera","name":"Lakera Guard","website":"https://www.lakera.ai","tagline":"Depot pick for a dedicated prompt-injection firewall in front of any LLM. NeMo when you need programmable…","description":"Depot pick for a dedicated prompt-injection firewall in front of any LLM. NeMo when you need programmable rails; OpenAI Moderation for free taxonomy flags. Check Point acquisition is real — product still has a public API. Strengths: Purpose-built prompt-injection / jailbreak / leakage API with sub-50ms narrative; Real free Community tier for eval; EU-native residency on the free plan.","categoryIds":["cat-observability","cat-inference-apis"],"jtbdIds":["prompt-injection-firewall","llm-input-guardrails","content-moderation"],"regions":["eu","us","global"],"pricingNotes":"As of 2026-09-14 (platform.lakera.ai/pricing): Community $0 — 10k requests/mo, 8k-token prompt cap, SaaS EU residency, SOC2/GDPR. Enterprise sales-gated — flexible volume, configurable prompt size, SaaS or self-host, SSO/RBAC/SIEM, EU/US residency. No public per-call overage. Acquired by Check Point (Sep 2025); enterprise quotes may route through Check Point. API: POST https://api.lakera.ai/v2/guard (OpenAI-style messages → flagged).","strengths":["Purpose-built prompt-injection / jailbreak / leakage API with sub-50ms narrative","Real free Community tier for eval","EU-native residency on the free plan"],"weaknesses":["Paid SKU is sales-gated (no Pro list price)","Post-Check Point packaging/quote path may slow indie procurement","Not a full policy engine like NeMo Guardrails"],"tags":["safety","prompt-injection","guardrails","checkpoint","firewall"],"scores":{"capabilityFit":8.8,"pricePerformance":7.8,"latencyReliability":8.4,"developerDx":9,"privacyPosture":8.8,"lockInRisk":5.5,"maturity":8,"overall":8.21,"kind":"curated_estimate","confidence":"high","asOf":"2026-09-14","sources":["Lakera Guard pricing: https://platform.lakera.ai/pricing","Lakera: https://www.lakera.ai","Prompt injection (Lakera): https://www.lakera.ai/risk/prompt-injection-attacks"],"notes":"Depot pick for a dedicated prompt-injection firewall in front of any LLM. NeMo when you need programmable rails; OpenAI Moderation for free taxonomy flags. Check Point acquisition is real — product still has a public API."},"lastReviewed":"2026-09-14","status":"active"},{"id":"prisma-airs","slug":"prisma-airs","name":"Palo Alto Prisma AIRS (ex-Protect AI)","website":"https://www.paloaltonetworks.com","tagline":"Catalog the current SKU (Prisma AIRS), not the dead Protect AI brand. Pick only if PANW is already the…","description":"Catalog the current SKU (Prisma AIRS), not the dead Protect AI brand. Pick only if PANW is already the security standard. Indie agents: Lakera + NeMo. Strengths: Broadest AI-lifecycle security platform (model scan → runtime → agent identity); Fits existing Palo Alto estates; Agent Gateway / Portkey lineage for production traffic control.","categoryIds":["cat-observability"],"jtbdIds":["prompt-injection-firewall","llm-input-guardrails"],"regions":["global","us"],"pricingNotes":"As of 2026-09-14: Protect AI is not independent — Palo Alto completed the acquisition 2025-07-22 and folded model scanning, posture, red teaming, runtime, and agent security into Prisma AIRS (3.0 added AI Agent Security / Agent Gateway preview). Enterprise platform SKU; no public per-call list price. Portkey acquisition (closed May 2026) adds a runtime AI gateway. Do not buy 'Protect AI' as a standalone vendor.","strengths":["Broadest AI-lifecycle security platform (model scan → runtime → agent identity)","Fits existing Palo Alto estates","Agent Gateway / Portkey lineage for production traffic control"],"weaknesses":["Sales-only; unusable as a credit-card agent default","Heavy platform lock-in vs a 20-line Lakera call","Independent Protect AI brand is gone"],"tags":["safety","enterprise","palo-alto","acquired","runtime"],"scores":{"capabilityFit":8.6,"pricePerformance":5,"latencyReliability":8.8,"developerDx":6.2,"privacyPosture":9,"lockInRisk":7.5,"maturity":9,"overall":7.28,"kind":"curated_estimate","confidence":"medium","asOf":"2026-09-14","sources":["Palo Alto completes Protect AI acquisition: https://www.paloaltonetworks.com/company/press/2025/palo-alto-networks-completes-acquisition-of-protect-ai","Prisma AIRS vs Cisco AI Defense 2026: https://www.kosmoy.com/resources/blog/cisco-ai-defense-vs-prisma-airs/"],"notes":"Catalog the current SKU (Prisma AIRS), not the dead Protect AI brand. Pick only if PANW is already the security standard. Indie agents: Lakera + NeMo."},"lastReviewed":"2026-09-14","status":"active"},{"id":"cisco-ai-defense","slug":"cisco-ai-defense","name":"Cisco AI Defense (ex-Robust Intelligence)","website":"https://www.cisco.com","tagline":"Status check: Robust Intelligence brand is gone; buy Cisco AI Defense only inside a Cisco security standard…","description":"Status check: Robust Intelligence brand is gone; buy Cisco AI Defense only inside a Cisco security standard. Lakera remains the self-serve firewall. Strengths: AI Firewall pioneer now inside Cisco Security Cloud traffic path; Fits Cisco networking/security estates; Lifecycle coverage from model risk to runtime.","categoryIds":["cat-observability"],"jtbdIds":["prompt-injection-firewall","llm-input-guardrails"],"regions":["global","us"],"pricingNotes":"As of 2026-09-14: Robust Intelligence is not independent — Cisco completed the acquisition 2024-09-24 and folded the AI Firewall / red-teaming stack into Cisco AI Defense + Cisco Foundation AI. Enterprise Security Cloud SKU; no public per-call list price. Astrix Security deal (reported ~$400M, May 2026) extends agent/identity surface. Do not buy 'Robust Intelligence' as a standalone vendor.","strengths":["AI Firewall pioneer now inside Cisco Security Cloud traffic path","Fits Cisco networking/security estates","Lifecycle coverage from model risk to runtime"],"weaknesses":["Sales-only; no indie rate card","Brand/SKU churn post-acquisition","Overkill vs Lakera for a single agent app"],"tags":["safety","enterprise","cisco","acquired","firewall"],"scores":{"capabilityFit":8.4,"pricePerformance":4.8,"latencyReliability":8.6,"developerDx":6,"privacyPosture":8.8,"lockInRisk":7.8,"maturity":8.8,"overall":7.07,"kind":"curated_estimate","confidence":"medium","asOf":"2026-09-14","sources":["Robust Intelligence is now part of Cisco: https://www.cisco.com/site/us/en/products/security/ai-defense/robust-intelligence-is-part-of-cisco/index.html","Cisco intent to acquire Robust Intelligence: https://blogs.cisco.com/news/fortifying-the-future-of-security-for-ai-cisco-announces-intent-to-acquire-robust-intelligence"],"notes":"Status check: Robust Intelligence brand is gone; buy Cisco AI Defense only inside a Cisco security standard. Lakera remains the self-serve firewall."},"lastReviewed":"2026-09-14","status":"active"},{"id":"cua","slug":"cua","name":"Cua","website":"https://cua.ai","tagline":"Best Scrapybara/E2B Desktop peer when you need Windows+macOS+Android fleets or local OSS sandboxes. E2B…","description":"Best Scrapybara/E2B Desktop peer when you need Windows+macOS+Android fleets or local OSS sandboxes. E2B remains the depot Linux-desktop default; Cua when OS coverage is the job. Strengths: True multi-OS computer-use fleets (Linux/Windows/macOS/Android) plus OSS local runtime; Warm pools / fork-from-snapshot for eval and RL rollouts; Background driver does not steal the user cursor.","categoryIds":["cat-sandboxes-browsers"],"jtbdIds":["desktop-computer-use","agent-sandbox","web-agent","mobile-computer-use"],"regions":["us","global"],"pricingNotes":"As of 2026-09-14 (cua.ai): OSS MIT sandbox/driver (Docker/QEMU/Apple VZ local). Cua Fleet usage-based: CPU $0.044625/vCPU/hr, memory $0.0223125/GB/hr. Cloud Windows sandboxes GA (credits: Small 8/hr, Medium 15/hr, Large 31/hr on the Windows GA post); macOS preview on patented virtualization. Linux + Android VMs. Warm pools, snapshots, BYOC/on-prem, SOC2 Type I.","strengths":["True multi-OS computer-use fleets (Linux/Windows/macOS/Android) plus OSS local runtime","Warm pools / fork-from-snapshot for eval and RL rollouts","Background driver does not steal the user cursor"],"weaknesses":["Younger hosted SLA story than E2B","Credit vs $/vCPU packaging can confuse forecasting","macOS still preview"],"tags":["desktop","computer-use","windows","macos","android","oss"],"scores":{"capabilityFit":8.8,"pricePerformance":8,"latencyReliability":7.6,"developerDx":8.6,"privacyPosture":8.2,"lockInRisk":3.8,"maturity":7,"overall":8.02,"kind":"curated_estimate","confidence":"high","asOf":"2026-09-14","sources":["Cua: https://cua.ai","Cua GitHub: https://github.com/trycua/cua","Cloud Windows GA + macOS preview: https://cua.ai/blog/cloud-windows-ga-macos-preview"],"notes":"Best Scrapybara/E2B Desktop peer when you need Windows+macOS+Android fleets or local OSS sandboxes. E2B remains the depot Linux-desktop default; Cua when OS coverage is the job."},"lastReviewed":"2026-09-14","status":"active"},{"id":"hyperbrowser","slug":"hyperbrowser","name":"Hyperbrowser","website":"https://www.hyperbrowser.ai","tagline":"Pick when you want hosted OpenAI/Claude computer-use on stealth Chromium without owning the browser fleet…","description":"Pick when you want hosted OpenAI/Claude computer-use on stealth Chromium without owning the browser fleet. Anchor still wins authenticated humanized profiles; Browserbase wins session polish. Strengths: Stealth Chromium + first-party OpenAI/Claude/Gemini/Meta/Grok computer-use APIs; Transparent credit math ($0.10 browser-hour is cheap vs Browserbase hours); Scrape/extract APIs in the same account.","categoryIds":["cat-sandboxes-browsers","cat-search-scrape"],"jtbdIds":["computer-use-browser","cloud-browser","web-agent","desktop-computer-use"],"regions":["global","us"],"pricingNotes":"As of 2026-09-14 (docs): 1 credit = $0.001. Browser usage 100 credits/hr ($0.10/hr); proxy 10k credits/GB ($10/GB). Fetch $0.001/page (+$0.009 with proxy); scrape $0.001/page; AI extract $30/M output tokens. HyperAgent / Browser Use steps $0.02/step. Hosted CUA token pass-through (e.g. computer-use-preview $3.15/$12.60 per 1M). Startup plan commonly $30/mo for 30k credits / 25 concurrent (confirm billing page). SOC2 Type II + HIPAA.","strengths":["Stealth Chromium + first-party OpenAI/Claude/Gemini/Meta/Grok computer-use APIs","Transparent credit math ($0.10 browser-hour is cheap vs Browserbase hours)","Scrape/extract APIs in the same account"],"weaknesses":["Credit forecasting is fiddly vs flat browser-hours","Not a full desktop OS (browser CUA, not Windows GUI)","Younger than Browserbase"],"tags":["browser","computer-use","stealth","credits"],"scores":{"capabilityFit":8.2,"pricePerformance":8.2,"latencyReliability":7.8,"developerDx":8.4,"privacyPosture":7.8,"lockInRisk":4.5,"maturity":6.8,"overall":7.8,"kind":"curated_estimate","confidence":"high","asOf":"2026-09-14","sources":["Hyperbrowser pricing docs: https://www.hyperbrowser.ai/docs/pricing","Hyperbrowser: https://www.hyperbrowser.ai","OpenAI CUA on Hyperbrowser: https://hyperbrowser.ai/docs/agents/openai-cua"],"notes":"Pick when you want hosted OpenAI/Claude computer-use on stealth Chromium without owning the browser fleet. Anchor still wins authenticated humanized profiles; Browserbase wins session polish."},"lastReviewed":"2026-09-14","status":"active"},{"id":"skyvern","slug":"skyvern","name":"Skyvern","website":"https://www.skyvern.com","tagline":"Pick for ops/RPA-replacement browser agents (invoices, portals, 2FA). Not a substitute for E2B/Cua desktops…","description":"Pick for ops/RPA-replacement browser agents (invoices, portals, 2FA). Not a substitute for E2B/Cua desktops. Browser Use Cloud if you want a general NL agent loop instead of workflow runs. Strengths: Vision-agent workflows that replace brittle RPA/XPath; OSS self-host + honest cloud credit table; CAPTCHA/2FA/credential vault integrations.","categoryIds":["cat-sandboxes-browsers","cat-search-scrape"],"jtbdIds":["browser-agent-loop","web-agent","computer-use-browser","form-filling-rpa"],"regions":["global","us"],"pricingNotes":"As of 2026-09-14 (skyvern.com/pricing): Free 5k credits (1 concurrent); Hobby $29/mo / 30k credits / 10 concurrent; Pro $149/mo / 150k credits / 25 concurrent; Enterprise custom (100 concurrent, HIPAA, SOC2, self-host). A run is one workflow execution (not per page/step); credits scale with complexity/duration. OSS self-host free (you pay LLM + browsers). 2FA/TOTP, 1Password, Azure Key Vault, Bitwarden on higher tiers.","strengths":["Vision-agent workflows that replace brittle RPA/XPath","OSS self-host + honest cloud credit table","CAPTCHA/2FA/credential vault integrations"],"weaknesses":["Workflow-platform, not a general desktop sandbox","Credit-per-run forecasting still workload-dependent","Less infra-control than raw Browserbase/Playwright"],"tags":["browser","rpa","vision","oss","workflow"],"scores":{"capabilityFit":8.4,"pricePerformance":8,"latencyReliability":7.8,"developerDx":8.2,"privacyPosture":8,"lockInRisk":4,"maturity":7.2,"overall":7.88,"kind":"curated_estimate","confidence":"high","asOf":"2026-09-14","sources":["Skyvern pricing: https://www.skyvern.com/pricing","Skyvern: https://www.skyvern.com"],"notes":"Pick for ops/RPA-replacement browser agents (invoices, portals, 2FA). Not a substitute for E2B/Cua desktops. Browser Use Cloud if you want a general NL agent loop instead of workflow runs."},"lastReviewed":"2026-09-14","status":"active"},{"id":"vapi","slug":"vapi","name":"Vapi","website":"https://vapi.ai","tagline":"Depot pick for shipping a phone/voice agent without assembling vendors. LiveKit when you want OSS + realtime…","description":"Depot pick for shipping a phone/voice agent without assembling vendors. LiveKit when you want OSS + realtime media control; Deepgram when STT quality is the product and you will orchestrate yourself. Strengths: The actual 'don't stitch STT+LLM+TTS' product with BYOK; Phone/SMS channels + concurrency SKU that maps to call centers; Clear $0.05/min hosting math.","categoryIds":["cat-comms","cat-media-gen"],"jtbdIds":["voice-agent-platform","voice-agents","speech-synthesis","speech-to-text"],"regions":["global","us"],"pricingNotes":"As of 2026-09-14 (vapi.ai/pricing): Build = usage-based hosting $0.05/min voice + $0.005/SMS/chat msg; STT/LLM/TTS at provider cost (or $0 with BYOK). 10 concurrent lines included, then $10/line/mo. Add-ons: HIPAA $2,000/mo; Zero Data Retention $1,000/mo. Scale = annual commit, volume $/min, SSO/RBAC/SOC2/PCI. Call history 14 days on Build.","strengths":["The actual 'don't stitch STT+LLM+TTS' product with BYOK","Phone/SMS channels + concurrency SKU that maps to call centers","Clear $0.05/min hosting math"],"weaknesses":["Hosting + model + telephony stack into $0.10–0.20/min all-in easily","HIPAA/ZDR add-ons are expensive vs LiveKit self-host","Less of a media-network primitive than LiveKit"],"tags":["voice","agents","telephony","stt","tts"],"scores":{"capabilityFit":9,"pricePerformance":7.2,"latencyReliability":8.4,"developerDx":9.2,"privacyPosture":7.5,"lockInRisk":5,"maturity":8,"overall":8.09,"kind":"curated_estimate","confidence":"high","asOf":"2026-09-14","sources":["Vapi pricing: https://vapi.ai/pricing","Vapi: https://vapi.ai"],"notes":"Depot pick for shipping a phone/voice agent without assembling vendors. LiveKit when you want OSS + realtime media control; Deepgram when STT quality is the product and you will orchestrate yourself."},"lastReviewed":"2026-09-14","status":"active"},{"id":"livekit","slug":"livekit","name":"LiveKit Cloud","website":"https://livekit.com","tagline":"Default when the agent is a realtime session (voice or video) you want to own. Vapi for turnkey telephony…","description":"Default when the agent is a realtime session (voice or video) you want to own. Vapi for turnkey telephony agents; Cartesia/Deepgram as leaves inside LiveKit Inference. Strengths: Best OSS+Cloud realtime media substrate for voice/video agents; Transparent per-min inference catalog; 99.99% network + region pinning / HIPAA on Scale.","categoryIds":["cat-comms","cat-media-gen"],"jtbdIds":["voice-agent-platform","voice-agents","realtime-tts"],"regions":["global","us"],"pricingNotes":"As of 2026-09-14 (livekit.com/pricing): Build $0 (1k agent minutes, $2.50 inference credits, 1 free US number); Ship from $50/mo (5k agent min then $0.01/min, $5 credits); Scale from $500/mo (50k min, $50 credits, HIPAA/region pin); Enterprise custom. Example all-in ~$0.067/min (LLM+STT+TTS+obs+$0.01 session). WebRTC $0.0004–$0.0005/min overage. OSS Agents framework + media server self-host free. Inference catalog includes Cartesia/Deepgram/AssemblyAI/Rime (some TTS SKUs free on LiveKit).","strengths":["Best OSS+Cloud realtime media substrate for voice/video agents","Transparent per-min inference catalog","99.99% network + region pinning / HIPAA on Scale"],"weaknesses":["You still compose STT/LLM/TTS (more assembly than Vapi)","Plan limits on concurrent sessions can surprise","Self-host ops burden if you skip Cloud"],"tags":["voice","webrtc","agents","oss","realtime"],"scores":{"capabilityFit":9.2,"pricePerformance":7.8,"latencyReliability":9,"developerDx":8.8,"privacyPosture":8.2,"lockInRisk":4.2,"maturity":8.8,"overall":8.48,"kind":"curated_estimate","confidence":"high","asOf":"2026-09-14","sources":["LiveKit pricing: https://livekit.com/pricing","LiveKit Inference pricing: https://livekit.com/pricing/inference","LiveKit: https://livekit.com"],"notes":"Default when the agent is a realtime session (voice or video) you want to own. Vapi for turnkey telephony agents; Cartesia/Deepgram as leaves inside LiveKit Inference."},"lastReviewed":"2026-09-14","status":"active"},{"id":"hume","slug":"hume","name":"Hume AI","website":"https://www.hume.ai","tagline":"A/B vs Cartesia when emotional prosody is the product. Cartesia remains the low-latency default; ElevenLabs…","description":"A/B vs Cartesia when emotional prosody is the product. Cartesia remains the low-latency default; ElevenLabs for library breadth. Strengths: Expressive / empathic TTS and speech-to-speech (EVI) differentiated vs Cartesia/ElevenLabs; Voice cloning + Octave 2; Enterprise compliance SKUs exist.","categoryIds":["cat-media-gen","cat-comms"],"jtbdIds":["speech-synthesis","realtime-tts","voice-agents"],"regions":["global","us"],"pricingNotes":"As of 2026-09-14 (hume.ai/pricing): Tiered Free → Creator → Team → Enterprise. EVI 3 additional usage $0.04/min (pinned). Team: 5 seats, 30 concurrent connections, unlimited voice create/use. Enterprise: custom characters/EVI, unlimited seats/concurrency, SOC2/GDPR/HIPAA. Included character bundles and Creator list $ are JS-rendered — verify in console; EVI unit price is solid.","strengths":["Expressive / empathic TTS and speech-to-speech (EVI) differentiated vs Cartesia/ElevenLabs","Voice cloning + Octave 2","Enterprise compliance SKUs exist"],"weaknesses":["Public $ table is harder to pin than Cartesia","EVI couples you to Hume's conversational stack","Not a telephony orchestrator"],"tags":["tts","evi","voice","emotion"],"scores":{"capabilityFit":8.6,"pricePerformance":7,"latencyReliability":7.8,"developerDx":8,"privacyPosture":7.8,"lockInRisk":5,"maturity":7.2,"overall":7.63,"kind":"curated_estimate","confidence":"high","asOf":"2026-09-14","sources":["Hume pricing: https://hume.ai/pricing","Hume AI: https://www.hume.ai"],"notes":"A/B vs Cartesia when emotional prosody is the product. Cartesia remains the low-latency default; ElevenLabs for library breadth."},"lastReviewed":"2026-09-14","status":"active"},{"id":"motherduck","slug":"motherduck","name":"MotherDuck","website":"https://motherduck.com","tagline":"Depot pick when the agent needs SQL warehouse context (analytics, DuckLake/Iceberg, shared Guides). Snowflake…","description":"Depot pick when the agent needs SQL warehouse context (analytics, DuckLake/Iceberg, shared Guides). Snowflake Cortex / Databricks Mosaic when the lakehouse is already the system of record. Strengths: DuckDB-shaped warehouse with first-class agent Guides/MCP/Flights; Honest Lite free tier + per-second compute; Dual execution (local DuckDB + cloud) cuts scan cost.","categoryIds":["cat-vector-memory","cat-cpu-compute"],"jtbdIds":["sql-warehouse-agent-context","serverless-db","embedded-vector","spreadsheet-sql-agents"],"regions":["us","eu","global"],"pricingNotes":"As of 2026-10-24 (motherduck.com/product/pricing, fetched 2026-10-24): Lite $0 — 3 users, 2 service accounts, 10 GB, 10 Pulse hrs/mo. Business $250/org/mo + usage — 10 users, unlimited service accounts, 99.9% SLA. Enterprise custom (PrivateLink, HIPAA BAA, capacity). Storage $0.04/GB-mo. Compute: Pulse $0.60/CU-hr (per query, 1 CU-sec min); Standard $2.40/hr; Jumbo $4.80; Mega $12; Giga $24 (per-second). Flights (agent-native pipelines), 'billed per second at the rate for the Flight's size': F4 '$0.15 per hour' ('Lite without a credit card includes 30 minutes per day at no charge.'), F16 '$0.60 per hour' (on Lite with a credit card), F32 '$1.20 per hour'; SQL a Flight runs is billed at the compute rate. AI functions $1.00/AI Unit. MCP + Guides for agent context. Regions: us-east-1, us-west-2, eu-central-1, eu-west-1, ap-northeast-1, ap-southeast-2 (one region per org).","strengths":["DuckDB-shaped warehouse with first-class agent Guides/MCP/Flights","Honest Lite free tier + per-second compute","Dual execution (local DuckDB + cloud) cuts scan cost"],"weaknesses":["Single-region orgs","Not a vector DB — VSS/extensions only","Business $250 base can dwarf tiny agent apps (use Lite)"],"tags":["duckdb","warehouse","sql","mcp","agents"],"scores":{"capabilityFit":8.8,"pricePerformance":8.6,"latencyReliability":8.6,"developerDx":9,"privacyPosture":8.5,"lockInRisk":4.5,"maturity":8.2,"overall":8.44,"kind":"curated_estimate","confidence":"high","asOf":"2026-10-19","sources":["MotherDuck pricing: https://motherduck.com/product/pricing/","MotherDuck billing docs: https://motherduck.com/docs/about-motherduck/billing/pricing/","MotherDuck: https://motherduck.com"],"notes":"Depot pick when the agent needs SQL warehouse context (analytics, DuckLake/Iceberg, shared Guides). Snowflake Cortex / Databricks Mosaic when the lakehouse is already the system of record."},"lastReviewed":"2026-10-19","status":"active"},{"id":"launchdarkly","slug":"launchdarkly","name":"LaunchDarkly","website":"https://launchdarkly.com","tagline":"Depot pick for progressive delivery of agent prompts/models/tools in production. Statsig if you want cheaper…","description":"Depot pick for progressive delivery of agent prompts/models/tools in production. Statsig if you want cheaper experimentation+flags in one product and can tolerate Amplitude-era packaging flux. Strengths: Purpose-built AgentControl for prompt/model/tool rollout + online evals; Mature targeting, approvals, and Guardian auto-rollback; No per-seat tax.","categoryIds":["cat-observability"],"jtbdIds":["progressive-agent-rollout","evals"],"regions":["global","us"],"pricingNotes":"As of 2026-09-14 (launchdarkly.com/pricing): Developer $0 — unlimited seats, 5 service connections, 1k client MAU, 5k AI runs/mo, 14-day AgentControl retention. Foundation PAYG (yearly): $10/service connection/mo + $8.33/1k client MAU/mo; 5k AI runs included then $5/1k; 30-day retention. Enterprise custom (approvals, SCIM, 100-day retention). Guardian add-on: guardrail metrics + auto pause/rollback. An AI Run = tracked LLM request or online LLM-as-judge eval. AgentControl configures prompts/models/tools and observes agent runs.","strengths":["Purpose-built AgentControl for prompt/model/tool rollout + online evals","Mature targeting, approvals, and Guardian auto-rollback","No per-seat tax"],"weaknesses":["Service-connection billing can explode on serverless/K8s (they will custom-quote)","More expensive than Statsig Pro's flat $150","Heavier than you need for a 2-person agent app"],"tags":["feature-flags","agents","experimentation","rollout"],"scores":{"capabilityFit":9,"pricePerformance":6.8,"latencyReliability":9.2,"developerDx":8.6,"privacyPosture":8.2,"lockInRisk":5.8,"maturity":9.4,"overall":8.2,"kind":"curated_estimate","confidence":"high","asOf":"2026-09-14","sources":["LaunchDarkly pricing: https://launchdarkly.com/pricing/","LaunchDarkly: https://launchdarkly.com"],"notes":"Depot pick for progressive delivery of agent prompts/models/tools in production. Statsig if you want cheaper experimentation+flags in one product and can tolerate Amplitude-era packaging flux."},"lastReviewed":"2026-09-14","status":"active"},{"id":"statsig","slug":"statsig","name":"Statsig","website":"https://www.statsig.com","tagline":"Best price/perf flags+experiments for product agents. Confirm Amplitude commercial terms at renewal…","description":"Best price/perf flags+experiments for product agents. Confirm Amplitude commercial terms at renewal. LaunchDarkly when you need AgentControl + Guardian rollback. Strengths: Flags + experiment stats + product analytics in one cheap SKU; Unlimited flag checks even on free; Warehouse-native path for serious data teams.","categoryIds":["cat-observability"],"jtbdIds":["progressive-agent-rollout","evals"],"regions":["global","us"],"pricingNotes":"As of 2026-09-14 (statsig.com/pricing): Developer free — 2M events/mo, unlimited flag/config checks, 50k replays, 1-yr analytics. Pro $150/mo — 5M events then $0.05/1k, 100k replays, unlimited retention, change reviews. Enterprise custom (warehouse-native, SSO/RBAC, HIPAA BAA). Ownership: Statsig team joined OpenAI (2025); product/brand/customers moved to Amplitude (Phase 1 2026) — statsig.com remains the live SKU but renewal packaging may evolve.","strengths":["Flags + experiment stats + product analytics in one cheap SKU","Unlimited flag checks even on free","Warehouse-native path for serious data teams"],"weaknesses":["Ownership/packaging churn (OpenAI team / Amplitude product)","Less agent-specific than LaunchDarkly AgentControl","Pro is per-project"],"tags":["feature-flags","experimentation","amplitude","openai"],"scores":{"capabilityFit":8.6,"pricePerformance":8.8,"latencyReliability":8.6,"developerDx":8.8,"privacyPosture":8,"lockInRisk":4.8,"maturity":8.2,"overall":8.33,"kind":"curated_estimate","confidence":"high","asOf":"2026-09-14","sources":["Statsig pricing: https://www.statsig.com/pricing","Statsig + Amplitude Phase 1: https://www.statsig.com/blog/statsig-amplitude-phase-1","Statsig is joining OpenAI: https://www.statsig.com/blog/openai-acquisition"],"notes":"Best price/perf flags+experiments for product agents. Confirm Amplitude commercial terms at renewal. LaunchDarkly when you need AgentControl + Guardian rollback."},"lastReviewed":"2026-09-14","status":"active"},{"id":"vantage","slug":"vantage","name":"Vantage","website":"https://www.vantage.sh","tagline":"Depot pick for unified cloud+GPU+LLM FinOps. Helicone/Langfuse still own request-path token tracing; Vantage…","description":"Depot pick for unified cloud+GPU+LLM FinOps. Helicone/Langfuse still own request-path token tracing; Vantage is the bill/allocation layer. CloudZero if you want a sales-led unit-economics platform. Strengths: Public fixed plans that do not scale as a % of spend; Native LLM + GPU + SaaS connectors (not just AWS bills); Virtual tagging for per-agent/per-customer allocation.","categoryIds":["cat-observability"],"jtbdIds":["agent-cloud-finops","cost-monitoring","llm-cost-anomaly-alerts"],"regions":["global","us"],"pricingNotes":"As of 2026-10-24 (vantage.sh/pricing, fetched 2026-10-24): 'Simple, Fixed Pricing'. Starter 'Free' with '$2,500 of cloud spend' and '3 Users'; Pro '$30 /month' with '$7,500 of cloud spend' and '5 Users'; Business '$200 /month' with '$20,000 of cloud spend' and '10 Users' (Pro and Business: '14 Day Free Trial'); Enterprise 'Custom' with 'Unlimited tracked cloud spend', 'Automated FinOps Agent' and 'Unlimited Users'. Data retention '6 Months' (Starter, Pro) and '12 Months' (Business). '40+ Supported Providers', including AI providers 'Anthropic Anyscale Baseten Cartesia Cursor ElevenLabs Fireworks AI Modal OpenAI OpenRouter'. The feature comparison lists 'LLM Token Allocation', 'Terraform Provider' and 'MCP Server for AI'; per-tier availability and any add-on price for them are not stated in the page text.","strengths":["Public fixed plans that do not scale as a % of spend","Native LLM + GPU + SaaS connectors (not just AWS bills)","Virtual tagging for per-agent/per-customer allocation"],"weaknesses":["Tracked-spend caps force plan jumps","Not a token gateway (Helicone still needed in the request path)","Enterprise FinOps Agent is sales-gated"],"tags":["finops","cost","llm","gpu","mcp"],"scores":{"capabilityFit":8.6,"pricePerformance":9,"latencyReliability":8.6,"developerDx":8.8,"privacyPosture":7.8,"lockInRisk":4,"maturity":8.4,"overall":8.42,"kind":"curated_estimate","confidence":"high","asOf":"2026-09-14","sources":["Vantage pricing: https://www.vantage.sh/pricing","Vantage LLM token allocation: https://www.vantage.sh/features/llm-token-allocation","Vantage: https://www.vantage.sh"],"notes":"Depot pick for unified cloud+GPU+LLM FinOps. Helicone/Langfuse still own request-path token tracing; Vantage is the bill/allocation layer. CloudZero if you want a sales-led unit-economics platform."},"lastReviewed":"2026-09-14","status":"active"},{"id":"cloudzero","slug":"cloudzero","name":"CloudZero","website":"https://www.cloudzero.com","tagline":"Pick when finance wants AI unit economics and will sit through a sales cycle. Vantage for self-serve agent…","description":"Pick when finance wants AI unit economics and will sit through a sales cycle. Vantage for self-serve agent teams. Strengths: Unit-economics / AI ROI narrative (cost per customer, per agent run); Unlimited users and sources once contracted; MCP for agent access to cost data.","categoryIds":["cat-observability"],"jtbdIds":["agent-cloud-finops","cost-monitoring"],"regions":["global","us"],"pricingNotes":"As of 2026-09-14: Single subscription, all capabilities included; custom quote aligned to environment complexity (not a public % of spend). Unlimited sources/users/dimensions/dashboards, hourly granularity, anomaly detection, MCP integration. No free tier / no list price. Positioned as an AI-economics control plane vs classic cloud bill tools.","strengths":["Unit-economics / AI ROI narrative (cost per customer, per agent run)","Unlimited users and sources once contracted","MCP for agent access to cost data"],"weaknesses":["No public rate card or free tier","Heavier sales cycle than Vantage's $0/$30/$200","Overkill for a single-product agent startup"],"tags":["finops","cost","enterprise","ai-roi"],"scores":{"capabilityFit":8.8,"pricePerformance":6.2,"latencyReliability":8.6,"developerDx":7.5,"privacyPosture":8.2,"lockInRisk":5.5,"maturity":8.6,"overall":7.73,"kind":"curated_estimate","confidence":"medium","asOf":"2026-09-14","sources":["CloudZero pricing: https://www.cloudzero.com/pricing/","CloudZero: https://www.cloudzero.com"],"notes":"Pick when finance wants AI unit economics and will sit through a sales cycle. Vantage for self-serve agent teams."},"lastReviewed":"2026-09-14","status":"active"},{"id":"upstash-qstash","slug":"upstash-qstash","name":"Upstash QStash","website":"https://upstash.com","tagline":"Depot pick for a raw serverless agent event backbone (webhooks, delayed retries, fan-out) without operating…","description":"Depot pick for a raw serverless agent event backbone (webhooks, delayed retries, fan-out) without operating Kafka/SQS. Inngest when you need durable steps; NATS/Synadia when you need a real messaging fabric; WarpStream/Confluent when you already are Kafka. Strengths: Serverless HTTP queue/scheduler with almost no ops — ideal agent callback bus; Penny-level PAYG and a real free tier; Works from Workers/Vercel/edge without a VPC.","categoryIds":["cat-cpu-compute"],"jtbdIds":["serverless-agent-queue","event-driven-agent-jobs","agent-backend","agent-dlq-retry"],"regions":["global","us"],"pricingNotes":"As of 2026-09-14 (upstash.com/pricing/qstash): Free 1k messages/day, 50 GB, 1 MB max. PAYG $1 per 100k messages (unlimited/day, 10 MB, 1-year delay); bandwidth free to 50 GB then $0.05/GB. Fixed 1M $180/mo; Fixed 10M $420/mo; Enterprise custom. Retries billed per delivery attempt. HTTP callbacks up to 2h (PAYG) / 12h (Fixed 10M). Prod Pack add-on for SOC2/SLA on non-Enterprise.","strengths":["Serverless HTTP queue/scheduler with almost no ops — ideal agent callback bus","Penny-level PAYG and a real free tier","Works from Workers/Vercel/edge without a VPC"],"weaknesses":["Not a workflow engine (no step state like Inngest/Temporal)","Retries are billable","SOC2 requires add-on until Enterprise"],"tags":["queue","serverless","scheduler","http","agents"],"scores":{"capabilityFit":8.2,"pricePerformance":9.2,"latencyReliability":8.4,"developerDx":9.2,"privacyPosture":7.8,"lockInRisk":3.5,"maturity":8,"overall":8.38,"kind":"curated_estimate","confidence":"high","asOf":"2026-09-14","sources":["Upstash QStash pricing: https://upstash.com/pricing/qstash","Upstash: https://upstash.com"],"notes":"Depot pick for a raw serverless agent event backbone (webhooks, delayed retries, fan-out) without operating Kafka/SQS. Inngest when you need durable steps; NATS/Synadia when you need a real messaging fabric; WarpStream/Confluent when you already are Kafka."},"lastReviewed":"2026-09-14","status":"active"},{"id":"synadia-cloud","slug":"synadia-cloud","name":"Synadia Cloud (NATS)","website":"https://www.synadia.com","tagline":"Pick when agents need a real pub/sub fabric (multi-service, edge leafs) not just HTTP callbacks. QStash for…","description":"Pick when agents need a real pub/sub fabric (multi-service, edge leafs) not just HTTP callbacks. QStash for simpler serverless; Kafka/WarpStream for log-shaped replay. Strengths: NATS/JetStream fabric with subject-based routing — excellent agent command/event bus; Cheap Starter/Pro vs Kafka clouds; Leaf nodes for edge/device agents.","categoryIds":["cat-cpu-compute"],"jtbdIds":["serverless-agent-queue","event-driven-agent-jobs"],"regions":["global","us"],"pricingNotes":"As of 2026-09-14 (docs.synadia.com/cloud/pricing): Personal free (2 accounts, 10 connections, 5 GiB, 10 GiB net). Starter $49/mo (3 accounts, 100 conn, 10 GiB std / 1 GiB HA). Pro $199/mo (10 accounts, 1k conn, 50 GiB std / 10 GiB HA). Enterprise custom + SLA. Overage on paid: client/leaf egress $0.09/GiB, gateway $0.02, route $0.01. Add-ons: remote system $49/mo; extra connections $10; HA storage $0.90/GiB-mo. NATS OSS self-host remains free.","strengths":["NATS/JetStream fabric with subject-based routing — excellent agent command/event bus","Cheap Starter/Pro vs Kafka clouds","Leaf nodes for edge/device agents"],"weaknesses":["You still write consumer logic (not a workflow product)","Personal plan drops connections on overage","Smaller ecosystem than Kafka/SQS"],"tags":["nats","queue","pubsub","jetstream","agents"],"scores":{"capabilityFit":8.6,"pricePerformance":8.2,"latencyReliability":8.8,"developerDx":8,"privacyPosture":8.5,"lockInRisk":3.8,"maturity":8.5,"overall":8.28,"kind":"curated_estimate","confidence":"high","asOf":"2026-09-14","sources":["Synadia Cloud pricing: https://docs.synadia.com/cloud/pricing","Synadia: https://www.synadia.com","NATS: https://nats.io"],"notes":"Pick when agents need a real pub/sub fabric (multi-service, edge leafs) not just HTTP callbacks. QStash for simpler serverless; Kafka/WarpStream for log-shaped replay."},"lastReviewed":"2026-09-14","status":"active"},{"id":"warpstream","slug":"warpstream","name":"WarpStream","website":"https://www.warpstream.com","tagline":"Pick when the agent backbone must be Kafka-protocol (replay, consumers, Iceberg) without Confluent prices…","description":"Pick when the agent backbone must be Kafka-protocol (replay, consumers, Iceberg) without Confluent prices. Confluent Cloud if you need full enterprise Kafka + connectors and already pay that tax. Strengths: Kafka-compat at a fraction of MSK/Confluent TCO (no cross-AZ replication tax); BYOC = data never leaves your account; No partition tax — agent topics can be chatty.","categoryIds":["cat-cpu-compute"],"jtbdIds":["serverless-agent-queue","event-driven-agent-jobs"],"regions":["global","us"],"pricingNotes":"As of 2026-09-14 (warpstream.com/pricing): Kafka-protocol BYOC. Writes uncompressed $0.01/GiB first 50 TiB/mo stepping to $0.003; storage $0.01–$0.004/GiB-mo; cluster minutes ~$0.0023/min (Dev ~$100/mo floor; Fundamentals $500; Pro $1,500; Enterprise custom). No per-partition / per-Agent / per-core fees. Idle clusters scale to $0. $400 non-expiring trial credits. Data plane in your cloud (object storage); control plane WarpStream. Tableflow extra for Iceberg. SOC2 T1/T2.","strengths":["Kafka-compat at a fraction of MSK/Confluent TCO (no cross-AZ replication tax)","BYOC = data never leaves your account","No partition tax — agent topics can be chatty"],"weaknesses":["BYOC still means you run Agents (less serverless than QStash)","Produce latency (~250ms median) is not NATS-class","Younger than Confluent for enterprise Kafka features"],"tags":["kafka","byoc","queue","streaming"],"scores":{"capabilityFit":8.8,"pricePerformance":9,"latencyReliability":8.6,"developerDx":7.8,"privacyPosture":9.2,"lockInRisk":4.2,"maturity":8,"overall":8.41,"kind":"curated_estimate","confidence":"high","asOf":"2026-09-14","sources":["WarpStream pricing: https://www.warpstream.com/pricing","WarpStream: https://www.warpstream.com"],"notes":"Pick when the agent backbone must be Kafka-protocol (replay, consumers, Iceberg) without Confluent prices. Confluent Cloud if you need full enterprise Kafka + connectors and already pay that tax."},"lastReviewed":"2026-09-14","status":"active"},{"id":"confluent-cloud","slug":"confluent-cloud","name":"Confluent Cloud","website":"https://www.confluent.io","tagline":"Enterprise Kafka default. Agent builders should prefer QStash (simple) or WarpStream (Kafka-compat cheaper)…","description":"Enterprise Kafka default. Agent builders should prefer QStash (simple) or WarpStream (Kafka-compat cheaper) unless Confluent is already standard. Strengths: Most complete managed Kafka + Flink + governance; Connector catalog for enterprise systems agents must talk to; Mature SLAs and multi-cloud.","categoryIds":["cat-cpu-compute"],"jtbdIds":["serverless-agent-queue","event-driven-agent-jobs"],"regions":["global","us"],"pricingNotes":"As of 2026-09-14 (confluent.io/pricing): Basic free-tier Kafka; Standard ~$0.75/eCKU-hr (commonly cited ~$385/mo entry); Enterprise from ~$895/mo; Freight from ~$2,300/mo. Plus throughput/storage/egress. Connectors, Flink, Stream Governance extra. This is the full managed Kafka suite — not a $1/100k HTTP queue. Confirm live calculator; packaging is SKU-dense.","strengths":["Most complete managed Kafka + Flink + governance","Connector catalog for enterprise systems agents must talk to","Mature SLAs and multi-cloud"],"weaknesses":["Expensive vs WarpStream/QStash for agent-scale buses","eCKU math is easy to under-estimate","Overkill if you only need delayed HTTP retries"],"tags":["kafka","streaming","enterprise","flink"],"scores":{"capabilityFit":9.2,"pricePerformance":5.8,"latencyReliability":9.4,"developerDx":7.5,"privacyPosture":8.5,"lockInRisk":5.5,"maturity":9.6,"overall":8,"kind":"curated_estimate","confidence":"high","asOf":"2026-09-14","sources":["Confluent pricing: https://www.confluent.io/pricing/","Confluent Cloud: https://www.confluent.io"],"notes":"Enterprise Kafka default. Agent builders should prefer QStash (simple) or WarpStream (Kafka-compat cheaper) unless Confluent is already standard."},"lastReviewed":"2026-09-14","status":"active"},{"id":"hashicorp-vault","slug":"hashicorp-vault","name":"HashiCorp Vault (HCP)","website":"https://www.hashicorp.com/products/vault","tagline":"Runner-up for agent secrets when HashiCorp is already the standard. Infisical remains the depot default for…","description":"Runner-up for agent secrets when HashiCorp is already the standard. Infisical remains the depot default for agent DX; Doppler for pure SaaS; AWS SM for IAM-native shops. Strengths: The enterprise secrets standard (dynamic creds, encryption as a service, many auth methods); HCP removes most ops; OSS avoids lock-in; MCP server for agent-native secret fetch.","categoryIds":["cat-auth-payments"],"jtbdIds":["agent-secrets"],"regions":["global","us"],"pricingNotes":"As of 2026-09-14: OSS Vault self-host free (you operate it). HCP Vault Dedicated is usage-based: cluster hours by edition/size (Starter/Development/Essentials/Standard × XS–L) billed per minute, plus monthly unique Clients. Secondary 2026 guides cite Essentials Small ~$1.58–$1.60/hr (~$1.1k/mo) + ~$72.92 per active client/mo; Development ~$0.03/hr with a 25-client cap. Confirm live HCP portal — HashiCorp does not publish a single marketing rate card. Official Vault MCP server exists (GitHub/Docker) for agent tool access.","strengths":["The enterprise secrets standard (dynamic creds, encryption as a service, many auth methods)","HCP removes most ops; OSS avoids lock-in","MCP server for agent-native secret fetch"],"weaknesses":["Client-based HCP bills surprise teams used to Infisical/Doppler seats","DX is operator-grade, not agent-builder-grade","No simple public starter SKU"],"tags":["secrets","vault","enterprise","mcp"],"scores":{"capabilityFit":9.2,"pricePerformance":6.2,"latencyReliability":9.4,"developerDx":7,"privacyPosture":9.5,"lockInRisk":5.5,"maturity":9.6,"overall":8.1,"kind":"curated_estimate","confidence":"high","asOf":"2026-09-14","sources":["HCP pricing definitions (Vault metrics): https://developer.hashicorp.com/hcp/docs/hcp/admin/billing/pricing-definitions","HCP Vault Dedicated tiers: https://developer.hashicorp.com/hcp/docs/vault/get-started/deployment-considerations/tiers-and-features","HashiCorp Vault: https://www.hashicorp.com/products/vault"],"notes":"Runner-up for agent secrets when HashiCorp is already the standard. Infisical remains the depot default for agent DX; Doppler for pure SaaS; AWS SM for IAM-native shops."},"lastReviewed":"2026-09-14","status":"active"},{"id":"glama","slug":"glama","name":"Glama","website":"https://glama.ai","tagline":"Directory/hosting MCP peer. The Official MCP Registry is the Depot pick for mcp-discovery since…","description":"Directory/hosting MCP peer. The Official MCP Registry is the Depot pick for mcp-discovery since 2026-10-24 (Smithery's connection services end 2026-10-31); Glama when you want gateway+hosted MCP slots with a public Free/$9/$26/$80 ladder. Strengths: Strong MCP directory/inspector UX and gateway; Free path to host/browse OSS MCP servers; Useful catalog peer so Smithery is not a monopoly.","categoryIds":["cat-search-scrape","cat-sandboxes-browsers"],"jtbdIds":["mcp-discovery","agent-tool-router"],"regions":["global","us"],"pricingNotes":"As of 2026-09-14: Official glama.ai/pricing publishes Free + Starter $9/mo ($9 AI credits, 3 hosted MCP servers, 100k logs/30d), Pro $26/mo ($26 credits, 10 servers, 90d logs), Business $80/mo ($80 credits, 30 servers, 180d logs); additional servers $4/$3/$2; OSS MCP hosting free. Unauthenticated scrapers may still 409/SPA-blank — tiers confirmed via first-party page content + secondary mirrors.","strengths":["Strong MCP directory/inspector UX and gateway","Free path to host/browse OSS MCP servers","Useful catalog peer so Smithery is not a monopoly"],"weaknesses":["SPA/bot scrapes of /pricing still flaky (409) even though tiers are published","Not a full tool-auth router like Composio/Nango","Younger than the official registry as a source of truth"],"tags":["mcp","registry","directory","gateway"],"scores":{"capabilityFit":7.8,"pricePerformance":8.5,"latencyReliability":7.2,"developerDx":8.2,"privacyPosture":7,"lockInRisk":4,"maturity":6.5,"overall":7.56,"kind":"curated_estimate","confidence":"high","asOf":"2026-09-14","sources":["Glama pricing: https://glama.ai/pricing","Glama MCP hosting: https://glama.ai/mcp/hosting","Glama: https://glama.ai"],"notes":"Directory/hosting MCP peer. The Official MCP Registry is the Depot pick for mcp-discovery since 2026-10-24 (Smithery's connection services end 2026-10-31); Glama when you want gateway+hosted MCP slots with a public Free/$9/$26/$80 ladder."},"lastReviewed":"2026-09-14","status":"active"},{"id":"pulse-mcp","slug":"pulse-mcp","name":"PulseMCP","website":"https://www.pulsemcp.com","tagline":"MCP directory peer to Glama/Smithery — best when you need a curated sub-registry for a platform, not…","description":"MCP directory peer to Glama/Smithery — best when you need a curated sub-registry for a platform, not self-serve hosting $. Official MCP Registry remains metadata source of truth. Strengths: Curated/live-tested MCP catalog with Official Registry ingest; Standards-aligned Sub-Registry API for gateways; Enrichment metrics beyond raw server.json.","categoryIds":["cat-search-scrape","cat-sandboxes-browsers"],"jtbdIds":["mcp-discovery","a2a-agent-discovery"],"regions":["global","us"],"pricingNotes":"As of 2026-09-14: Public MCP directory browse free. Commercial Managed MCP Sub-Registry/Catalog + Registry API are partnership/quote-only (hello@pulsemcp.com) — no published plan table (APIs.io confirms deliberate contact-us). Sub-Registry API implements Generic MCP Registry API + Pulse enrichments (popularity, security, authOptions premium).","strengths":["Curated/live-tested MCP catalog with Official Registry ingest","Standards-aligned Sub-Registry API for gateways","Enrichment metrics beyond raw server.json"],"weaknesses":["Commercial SKU sales-gated","Not a hosting runtime like Glama","Indie builders get directory value only"],"tags":["mcp","registry","directory","a2a-adjacent"],"scores":{"capabilityFit":8.4,"pricePerformance":7,"latencyReliability":7.8,"developerDx":8.2,"privacyPosture":7.5,"lockInRisk":4,"maturity":7.2,"overall":7.65,"kind":"curated_estimate","confidence":"medium","asOf":"2026-09-14","sources":["PulseMCP API: https://www.pulsemcp.com/api","PulseMCP Sub-Registry docs: https://www.pulsemcp.com/api/docs/v0.1","PulseMCP work with us: https://www.pulsemcp.com/work-with-us"],"notes":"MCP directory peer to Glama/Smithery — best when you need a curated sub-registry for a platform, not self-serve hosting $. Official MCP Registry remains metadata source of truth."},"lastReviewed":"2026-09-14","status":"active"},{"id":"aws-agent-registry","slug":"aws-agent-registry","name":"AWS Agent Registry","website":"https://aws.amazon.com/bedrock/agentcore/","tagline":"Depot pick for enterprise A2A/MCP discovery when the fleet already lives on AWS. OpenAgora/a2a-registry.dev…","description":"Depot pick for enterprise A2A/MCP discovery when the fleet already lives on AWS. OpenAgora/a2a-registry.dev for public/OSS discovery; PulseMCP/Smithery for MCP-only directories. Strengths: Native A2A + MCP enterprise catalog with approval workflows; Clear consumption Free Tier; Fits AWS AgentCore / IAM estates.","categoryIds":["cat-search-scrape","cat-sandboxes-browsers"],"jtbdIds":["a2a-agent-discovery","mcp-discovery"],"regions":["us","eu","global"],"pricingNotes":"As of 2026-09-14 (aws.amazon.com/bedrock/agentcore/pricing): AWS Agent Registry — first 5k records/mo free, then $0.40/1k records; Search API first 1M free then $0.020/1k; List+Get first 2M free then $0.004/1k. Catalogs MCP servers, A2A agent cards, skills, custom JSON. Regions include us-east-1, us-west-2, eu-west-1, ap-northeast-1, ap-southeast-2.","strengths":["Native A2A + MCP enterprise catalog with approval workflows","Clear consumption Free Tier","Fits AWS AgentCore / IAM estates"],"weaknesses":["AWS lock-in for cross-cloud fleets","Not a public internet directory","Only useful if already on AgentCore/Bedrock path"],"tags":["a2a","mcp","aws","agentcore","registry"],"scores":{"capabilityFit":8.6,"pricePerformance":8,"latencyReliability":8.8,"developerDx":7.8,"privacyPosture":8.2,"lockInRisk":7,"maturity":8.5,"overall":7.96,"kind":"curated_estimate","confidence":"high","asOf":"2026-09-14","sources":["AgentCore pricing (incl. Registry): https://aws.amazon.com/bedrock/agentcore/pricing/","AWS Agent Registry blog: https://aws.amazon.com/blogs/machine-learning/manage-agents-tools-and-skills-at-scale-with-aws-agent-registry/"],"notes":"Depot pick for enterprise A2A/MCP discovery when the fleet already lives on AWS. OpenAgora/a2a-registry.dev for public/OSS discovery; PulseMCP/Smithery for MCP-only directories."},"lastReviewed":"2026-09-14","status":"active"},{"id":"google-agent-registry","slug":"google-agent-registry","name":"Google Cloud Agent Registry","website":"https://docs.cloud.google.com/agent-registry/overview","tagline":"GCP twin of AWS Agent Registry. Use when agents already sit on Vertex/Gemini Enterprise; otherwise prefer…","description":"GCP twin of AWS Agent Registry. Use when agents already sit on Vertex/Gemini Enterprise; otherwise prefer open A2A registries. Strengths: First-party A2A+MCP governance inside GCP; Agent Card indexing for A2A skills; Fits Gemini Enterprise Agent Platform.","categoryIds":["cat-search-scrape","cat-sandboxes-browsers"],"jtbdIds":["a2a-agent-discovery","mcp-discovery"],"regions":["global","us","eu"],"pricingNotes":"As of 2026-09-14: Part of Gemini Enterprise Agent Platform. Skill Registry storage billed as Agent Storage $0.30/GiB-month (billing commenced 2026-09-01); read ops 1 Agent Compute vCPU-h ($0.085) per 3M reads; write 1 vCPU-h per 1M writes. Supports A2A 0.3/1.0 Agent Cards + MCP servers/skills. Auto-registers supported GCP runtimes in-project.","strengths":["First-party A2A+MCP governance inside GCP","Agent Card indexing for A2A skills","Fits Gemini Enterprise Agent Platform"],"weaknesses":["GCP estate lock-in","Pricing buried in Agent Platform SKUs","Not a public multi-vendor directory"],"tags":["a2a","mcp","gcp","registry","gemini"],"scores":{"capabilityFit":8.5,"pricePerformance":7.2,"latencyReliability":8.8,"developerDx":7.6,"privacyPosture":8.2,"lockInRisk":7.2,"maturity":8.8,"overall":7.78,"kind":"curated_estimate","confidence":"high","asOf":"2026-09-14","sources":["Agent Registry overview: https://docs.cloud.google.com/agent-registry/overview","Gemini Enterprise Agent Platform pricing: https://cloud.google.com/products/gemini-enterprise-agent-platform/pricing","Register agents (A2A): https://docs.cloud.google.com/agent-registry/register-agents"],"notes":"GCP twin of AWS Agent Registry. Use when agents already sit on Vertex/Gemini Enterprise; otherwise prefer open A2A registries."},"lastReviewed":"2026-09-14","status":"active"},{"id":"a2a-registry","slug":"a2a-registry","name":"A2A Registry (a2a-registry.dev)","website":"https://a2a-registry.dev/","tagline":"Lightweight public/self-host A2A discovery. Prefer OpenAgora if you want LF-aligned A2A v1.0 + live testing…","description":"Lightweight public/self-host A2A discovery. Prefer OpenAgora if you want LF-aligned A2A v1.0 + live testing; AWS/GCP registries for enterprise governance. Strengths: Purpose-built A2A v0.3 hosted registry; JSON-RPC + REST + self-host option; Low friction for protocol experiments.","categoryIds":["cat-search-scrape"],"jtbdIds":["a2a-agent-discovery"],"regions":["global"],"pricingNotes":"As of 2026-09-14: Hosted production registry at registry.a2a-registry.dev / api.a2a-registry.dev (JSON-RPC primary, REST alt). A2A Protocol v0.3.0. pip install a2a-registry for self-host. No published commercial rate card — treat hosted as free/community unless a paid tier appears.","strengths":["Purpose-built A2A v0.3 hosted registry","JSON-RPC + REST + self-host option","Low friction for protocol experiments"],"weaknesses":["Early/community maturity vs hyperscaler registries","No enterprise SLA/price clarity","Ecosystem still thinner than MCP directories"],"tags":["a2a","registry","oss","json-rpc"],"scores":{"capabilityFit":7.8,"pricePerformance":9,"latencyReliability":7,"developerDx":8,"privacyPosture":7,"lockInRisk":2.5,"maturity":5.5,"overall":7.6,"kind":"curated_estimate","confidence":"medium","asOf":"2026-09-14","sources":["A2A Registry: https://a2a-registry.dev/"],"notes":"Lightweight public/self-host A2A discovery. Prefer OpenAgora if you want LF-aligned A2A v1.0 + live testing; AWS/GCP registries for enterprise governance."},"lastReviewed":"2026-09-14","status":"active"},{"id":"openagora","slug":"openagora","name":"OpenAgora","website":"https://openagora.cc","tagline":"Best open A2A discovery default for multi-vendor agents. AgentVerse if you are already Fetch.ai/uAgents…","description":"Best open A2A discovery default for multi-vendor agents. AgentVerse if you are already Fetch.ai/uAgents; hyperscaler registries if governance/IAM is the product. Strengths: A2A v1.0 / LF AAIF alignment; Live task testing + Trust Gateway; Self-host + free public discovery.","categoryIds":["cat-search-scrape"],"jtbdIds":["a2a-agent-discovery"],"regions":["global"],"pricingNotes":"As of 2026-09-14: Open-source (Apache 2.0) public A2A v1.0 agent registry — free registration via web/API. Trust Gateway (HMAC), live test panel, x402/MPP payment hooks. Self-host full stack. No commercial seat/meter table (public infrastructure posture).","strengths":["A2A v1.0 / LF AAIF alignment","Live task testing + Trust Gateway","Self-host + free public discovery"],"weaknesses":["Younger than AgentVerse scale","No enterprise support SKU yet","Payment rails still early"],"tags":["a2a","registry","oss","linux-foundation"],"scores":{"capabilityFit":8.2,"pricePerformance":9.2,"latencyReliability":7.2,"developerDx":8.4,"privacyPosture":7.5,"lockInRisk":2,"maturity":5.8,"overall":7.94,"kind":"curated_estimate","confidence":"medium","asOf":"2026-09-14","sources":["OpenAgora docs: https://openagora.cc/docs/welcome-to-openagora","OpenAgora vs AgentVerse: https://openagora.cc/docs/openagora-vs-agentverse-choosing-an-ai-agent-registry-in-2026","OpenAgora GitHub: https://github.com/Noxr3/openagora"],"notes":"Best open A2A discovery default for multi-vendor agents. AgentVerse if you are already Fetch.ai/uAgents; hyperscaler registries if governance/IAM is the product."},"lastReviewed":"2026-09-14","status":"active"},{"id":"ionos-ai","slug":"ionos-ai","name":"IONOS AI Model Hub","website":"https://cloud.ionos.com/managed/ai-model-hub","tagline":"Strong Germany-sovereign open-model peer beside Scaleway/OVH/STACKIT. Mistral still default EU brand for…","description":"Strong Germany-sovereign open-model peer beside Scaleway/OVH/STACKIT. Mistral still default EU brand for agents; IONOS when DE residency + public € card matter. Strengths: Published EU € token matrix; German sovereign hosting story; OpenAI-compat + image/embeddings.","categoryIds":["cat-inference-apis"],"jtbdIds":["eu-sovereign-llm","open-model-inference","budget-open-inference"],"regions":["eu"],"pricingNotes":"As of 2026-09-14 (official EUR price list v2026-08-27): PAYG €/1M tokens — Llama 3.1 8B €0.15/€0.15; Mistral Nemo €0.15/€0.15; Mistral Small 24B €0.10/€0.30; gpt-oss-120b €0.15/€0.65; Llama 3.3 70B €0.65/€0.65; Llama 3.1 405B €1.75/€1.75; Qwen3.5-9B €0.10/€0.15; Qwen3-Coder-Next €0.15/€0.80; embeddings from €0.01. FLUX.1 schnell €0.0288/image. DE ISO 27001 data centers; OpenAI-compatible API. H200 GPU VMs from €3.00/hr (1×).","strengths":["Published EU € token matrix","German sovereign hosting story","OpenAI-compat + image/embeddings"],"weaknesses":["Frontier closed models absent","Smaller US DX mindshare than Mistral/Scaleway","Commercial-use framing"],"tags":["eu","sovereign","openai-compat","germany"],"scores":{"capabilityFit":7.8,"pricePerformance":8.4,"latencyReliability":8,"developerDx":7.8,"privacyPosture":9,"lockInRisk":3.5,"maturity":8.2,"overall":8.01,"kind":"curated_estimate","confidence":"high","asOf":"2026-09-14","sources":["IONOS Cloud EUR price list: https://docs.ionos.com/cloud/support/general-information/price-list/ionos-cloud-eur-en.md","IONOS AI Model Hub: https://cloud.ionos.com/managed/ai-model-hub","AI Model Hub docs: https://docs.ionos.com/cloud/ai/ai-model-hub"],"notes":"Strong Germany-sovereign open-model peer beside Scaleway/OVH/STACKIT. Mistral still default EU brand for agents; IONOS when DE residency + public € card matter."},"lastReviewed":"2026-09-14","status":"active"},{"id":"t-systems-aifs","slug":"t-systems-aifs","name":"T-Systems AI Foundation Services","website":"https://docs.llmhub.t-systems.net/","tagline":"Enterprise EU sovereign runner when Telekom procurement wins RFPs. Not a Depot default for startups — use…","description":"Enterprise EU sovereign runner when Telekom procurement wins RFPs. Not a Depot default for startups — use Mistral/Scaleway/IONOS/OVH instead. Strengths: Telekom sovereign T-Cloud + external hosts; Clear minimum-commit rate plans; Enterprise DE/EU procurement path.","categoryIds":["cat-inference-apis"],"jtbdIds":["eu-sovereign-llm","enterprise-llm"],"regions":["eu"],"pricingNotes":"As of 2026-09-14 (docs.llmhub.t-systems.net/plans, rates valid from 2026-08-01): Shared LLM Serving — Essential €1,000/mo min, Professional €3,000, Agentic €5,000, Enterprise custom. Example T-Cloud LLM €0.20/€0.65 per 1M in/out; cached input 15% of input. RPM ceilings 300/600/1000 best-effort (not SLA). Dedicated LLM Serving quote-only for private/fine-tuned/SLA latency.","strengths":["Telekom sovereign T-Cloud + external hosts","Clear minimum-commit rate plans","Enterprise DE/EU procurement path"],"weaknesses":["€1k floor kills indie agents","Shared RPM/TPM best-effort only","Less self-serve DX than Mistral/Scaleway"],"tags":["eu","sovereign","telekom","enterprise"],"scores":{"capabilityFit":8,"pricePerformance":5.5,"latencyReliability":8.2,"developerDx":6.8,"privacyPosture":9.2,"lockInRisk":5.5,"maturity":8.8,"overall":7.36,"kind":"curated_estimate","confidence":"high","asOf":"2026-09-14","sources":["AIFS plans & pricing: https://docs.llmhub.t-systems.net/plans/","AIFS LLMs: https://docs.llmhub.t-systems.net/models/llms/"],"notes":"Enterprise EU sovereign runner when Telekom procurement wins RFPs. Not a Depot default for startups — use Mistral/Scaleway/IONOS/OVH instead."},"lastReviewed":"2026-09-14","status":"active"},{"id":"guardrails-ai","slug":"guardrails-ai","name":"Guardrails AI","website":"https://guardrailsai.com","tagline":"OSS programmable rails peer to NeMo Guardrails. Lakera when you want a hosted injection firewall API…","description":"OSS programmable rails peer to NeMo Guardrails. Lakera when you want a hosted injection firewall API; Guardrails Hub when validators-as-code is the product. Strengths: Largest OSS validator Hub; Local/self-host path with no meter tax; Composable with any LLM.","categoryIds":["cat-observability","cat-inference-apis"],"jtbdIds":["llm-input-guardrails","content-moderation","prompt-injection-firewall"],"regions":["global","us"],"pricingNotes":"As of 2026-09-14: OSS Guardrails + Guardrails Hub (70+ validators) free/self-host. Guardrails Pro enterprise managed — usage-based / quote; AWS Marketplace listings historically cite ~$50k annual optional contracts. No public per-call Pro matrix — treat Pro as sales-led.","strengths":["Largest OSS validator Hub","Local/self-host path with no meter tax","Composable with any LLM"],"weaknesses":["Pro packaging opaque","You own latency/ops when self-hosting","Less turnkey firewall than Lakera"],"tags":["safety","guardrails","oss","validators"],"scores":{"capabilityFit":8.4,"pricePerformance":8.8,"latencyReliability":7.8,"developerDx":8.2,"privacyPosture":8.5,"lockInRisk":3,"maturity":7.8,"overall":8.2,"kind":"curated_estimate","confidence":"medium","asOf":"2026-09-14","sources":["Guardrails AI: https://guardrailsai.com","Guardrails Pro announcement: https://guardrailsai.com/blog/guardrails-pro"],"notes":"OSS programmable rails peer to NeMo Guardrails. Lakera when you want a hosted injection firewall API; Guardrails Hub when validators-as-code is the product."},"lastReviewed":"2026-09-14","status":"active"},{"id":"arthur-ai","slug":"arthur-ai","name":"Arthur AI","website":"https://www.arthur.ai","tagline":"Monitoring/evals peer with a real $60 Premium. Prefer Braintrust/Langfuse for tracing defaults; Arthur when…","description":"Monitoring/evals peer with a real $60 Premium. Prefer Braintrust/Langfuse for tracing defaults; Arthur when governance+Shield narrative matters. Strengths: Public Free/Premium monitoring+evals card; Agent-oriented spans/jobs metering; Shield firewall for enterprise.","categoryIds":["cat-observability"],"jtbdIds":["production-guardrail-evals","llm-tracing","llm-input-guardrails","runtime-llm-firewall"],"regions":["global","us"],"pricingNotes":"As of 2026-09-14 (arthur.ai/pricing): Free $0 — 4 use cases, 5k jobs / 300k spans / 12k inferences / 3k evals, 7-day retention. Premium $60/mo — 100 use cases, 20k jobs / 1.2M spans / 100k inferences / 75k evals, 30-day retention, custom dashboards/alerts. Enterprise custom (VPC, SSO, BAA, SLA). Arthur Shield LLM firewall is separately sales-gated.","strengths":["Public Free/Premium monitoring+evals card","Agent-oriented spans/jobs metering","Shield firewall for enterprise"],"weaknesses":["Shield price opaque","Not a specialist injection API like Lakera","Premium limits still thin vs Braintrust scale"],"tags":["safety","evals","monitoring","firewall"],"scores":{"capabilityFit":8.2,"pricePerformance":8,"latencyReliability":8,"developerDx":8.2,"privacyPosture":7.8,"lockInRisk":4.5,"maturity":8,"overall":7.89,"kind":"curated_estimate","confidence":"high","asOf":"2026-09-14","sources":["Arthur pricing: https://www.arthur.ai/pricing","Arthur Shield: https://www.arthur.ai/product/shield"],"notes":"Monitoring/evals peer with a real $60 Premium. Prefer Braintrust/Langfuse for tracing defaults; Arthur when governance+Shield narrative matters."},"lastReviewed":"2026-09-14","status":"active"},{"id":"hiddenlayer","slug":"hiddenlayer","name":"HiddenLayer","website":"https://www.hiddenlayer.com","tagline":"Enterprise AI security peer to Prisma AIRS / Cisco AI Defense. Lakera/NeMo/Guardrails for builders…","description":"Enterprise AI security peer to Prisma AIRS / Cisco AI Defense. Lakera/NeMo/Guardrails for builders; HiddenLayer for security-buyer RFPs. Strengths: Broad AI security (not just prompts); Enterprise SOC narrative; Runtime traffic inspection.","categoryIds":["cat-observability"],"jtbdIds":["llm-input-guardrails","prompt-injection-firewall"],"regions":["global","us"],"pricingNotes":"As of 2026-09-14: Enterprise AI security platform (model + supply-chain + runtime guardrails). No public list price; secondary estimates ~$30k–$150k/yr. Sales-led only — skip unless security RFP already shortlists them.","strengths":["Broad AI security (not just prompts)","Enterprise SOC narrative","Runtime traffic inspection"],"weaknesses":["Fully opaque pricing","Overkill for indie agents","Procurement-heavy"],"tags":["safety","enterprise","model-security"],"scores":{"capabilityFit":8.4,"pricePerformance":4,"latencyReliability":8.2,"developerDx":6,"privacyPosture":8.8,"lockInRisk":6.5,"maturity":8,"overall":6.88,"kind":"curated_estimate","confidence":"medium","asOf":"2026-09-14","sources":["HiddenLayer: https://www.hiddenlayer.com"],"notes":"Enterprise AI security peer to Prisma AIRS / Cisco AI Defense. Lakera/NeMo/Guardrails for builders; HiddenLayer for security-buyer RFPs."},"lastReviewed":"2026-09-14","status":"active"},{"id":"metronome","slug":"metronome","name":"Metronome","website":"https://metronome.com","tagline":"Strong usage-billing peer when Orb is too sales-opaque and you need SQL metrics + commits. OpenMeter/Kong for…","description":"Strong usage-billing peer when Orb is too sales-opaque and you need SQL metrics + commits. OpenMeter/Kong for simpler metering; Lago for OSS; Autumn for Stripe entitlements DX. Strengths: SQL metrics + real-time spend controls; Enterprise contract/commit primitives; Strong Stripe path.","categoryIds":["cat-auth-payments"],"jtbdIds":["usage-metering","billing"],"regions":["global","us"],"pricingNotes":"As of 2026-09-14 (metronome.com/pricing): Startup — billing volume at 0.8% + events at $0.04/1k ingest; SQL billable metrics, commits/credits, Stripe-native. Custom — Salesforce/NetSuite/cloud marketplace invoicing, warehouse exports, dedicated AM/SLA. Marketing also cites Starter economics around $100k billing volume / 10M events before %/event overages — confirm contract.","strengths":["SQL metrics + real-time spend controls","Enterprise contract/commit primitives","Strong Stripe path"],"weaknesses":["Percent-of-billing can get expensive at scale","Heavier than Lago OSS","Startup→Custom jump is sales"],"tags":["billing","metering","usage","stripe"],"scores":{"capabilityFit":9,"pricePerformance":7,"latencyReliability":8.6,"developerDx":8.4,"privacyPosture":7.8,"lockInRisk":5.5,"maturity":8.4,"overall":8,"kind":"curated_estimate","confidence":"high","asOf":"2026-09-14","sources":["Metronome pricing: https://metronome.com/pricing","Metronome: https://metronome.com"],"notes":"Strong usage-billing peer when Orb is too sales-opaque and you need SQL metrics + commits. OpenMeter/Kong for simpler metering; Lago for OSS; Autumn for Stripe entitlements DX."},"lastReviewed":"2026-09-14","status":"active"},{"id":"autumn","slug":"autumn","name":"Autumn","website":"https://useautumn.com","tagline":"Best DX when you already run Stripe and need synchronous feature gates for AI usage. Metronome/Orb for…","description":"Best DX when you already run Stripe and need synchronous feature gates for AI usage. Metronome/Orb for pricing science; Lago/OpenMeter for pure meters. Strengths: Tiny API surface for agent gating (check/track); OSS + Stripe-native; Entitlements not just invoices.","categoryIds":["cat-auth-payments"],"jtbdIds":["usage-metering","billing"],"regions":["global","us"],"pricingNotes":"As of 2026-09-14: Open-source billing infrastructure on top of Stripe — attach/check/track API for subscriptions, usage, credits, entitlements. Stripe still processes payments. Self-host OSS; hosted packaging is productized for AI startups — confirm current hosted $ on docs/dashboard (no classic seat table on homepage).","strengths":["Tiny API surface for agent gating (check/track)","OSS + Stripe-native","Entitlements not just invoices"],"weaknesses":["Not a full quote-to-cash suite like Orb/Metronome","Hosted price page thin","Requires Stripe"],"tags":["billing","stripe","oss","entitlements","ai-startups"],"scores":{"capabilityFit":8,"pricePerformance":8.8,"latencyReliability":7.8,"developerDx":9.2,"privacyPosture":7.5,"lockInRisk":4,"maturity":6.5,"overall":7.95,"kind":"curated_estimate","confidence":"medium","asOf":"2026-09-14","sources":["Autumn: https://useautumn.com","Autumn docs: https://docs.useautumn.com/welcome"],"notes":"Best DX when you already run Stripe and need synchronous feature gates for AI usage. Metronome/Orb for pricing science; Lago/OpenMeter for pure meters."},"lastReviewed":"2026-09-14","status":"active"},{"id":"amberflo","slug":"amberflo","name":"Amberflo","website":"https://www.amberflo.io","tagline":"Metering peer when you care about per-customer AI margin, not just invoices. Vantage/CloudZero for cloud…","description":"Metering peer when you care about per-customer AI margin, not just invoices. Vantage/CloudZero for cloud FinOps; OpenMeter for simple meters. Strengths: Cost-to-serve / margin visibility for AI products; Metering + billing in one; Useful when COGS attribution is the pain.","categoryIds":["cat-auth-payments"],"jtbdIds":["usage-metering","billing","agent-cloud-finops"],"regions":["global","us"],"pricingNotes":"As of 2026-09-14: Usage metering + customer margin/cost attribution for AI/infra. Marketing shows ~$99 and ~$599/mo plan bands + usage overages — treat as directional; confirm console. As of 2026-10-25 https://www.amberflo.io/pricing could not be re-checked: two plain-curl fetches about 10 s apart both failed the TLS handshake (curl error 35, 'tls alert handshake failure', no HTTP status), and https://amberflo.io/pricing failed the same way. The ~$99/~$599 bands above are not re-confirmed and not re-priced.","strengths":["Cost-to-serve / margin visibility for AI products","Metering + billing in one","Useful when COGS attribution is the pain"],"weaknesses":["https://www.amberflo.io/pricing failed the TLS handshake to plain curl (fetched twice 2026-10-25, no HTTP status), so plan prices are unconfirmed","Less mindshare than Orb/Metronome","May overlap Vantage for pure FinOps"],"tags":["billing","metering","finops","margin"],"scores":{"capabilityFit":7.8,"pricePerformance":7.2,"latencyReliability":7.6,"developerDx":7.8,"privacyPosture":7.5,"lockInRisk":5,"maturity":7.2,"overall":7.38,"kind":"curated_estimate","confidence":"medium","asOf":"2026-09-14","sources":["Amberflo pricing: https://amberflo.io/pricing","Amberflo: https://www.amberflo.io"],"notes":"Metering peer when you care about per-customer AI margin, not just invoices. Vantage/CloudZero for cloud FinOps; OpenMeter for simple meters."},"lastReviewed":"2026-09-14","status":"active"},{"id":"zoominfo","slug":"zoominfo","name":"ZoomInfo GTM AI","website":"https://gtm.ai","tagline":"Strong enrichment peer when agents need MCP + verified contacts. PDL remains Depot pick for API-first…","description":"Strong enrichment peer when agents need MCP + verified contacts. PDL remains Depot pick for API-first $/record research pipelines; Apollo for GTM suites; ZoomInfo when quality+MCP matter more than unit price. Strengths: Agent-native MCP enrichment with free search; Verified DB vs scrape; Transparent PAYG after free credits.","categoryIds":["cat-search-scrape"],"jtbdIds":["people-enrichment"],"regions":["global","us"],"pricingNotes":"As of 2026-09-14 (gtm.ai/pricing): Free — 1k data + 1k AI credits, MCP+API+CLI, no seats. PAYG top-ups from $20; Data credits ~$0.38 at $1k spend tier (volume discounts); AI credits $0.05 flat. Search/lookup free; enrich 1 credit/new record; Enterprise adds 12-mo Records Under Management, SSO, waterfall. MCP included with ZoomInfo subscriptions against same credit pools.","strengths":["Agent-native MCP enrichment with free search","Verified DB vs scrape","Transparent PAYG after free credits"],"weaknesses":["Data credit ~$0.38 list is steep vs PDL at volume","RUM benefit gated to Enterprise","GTM-skewed vs pure research APIs"],"tags":["enrichment","mcp","people","company","gtm"],"scores":{"capabilityFit":8.8,"pricePerformance":7,"latencyReliability":8.6,"developerDx":8.8,"privacyPosture":6.2,"lockInRisk":4.5,"maturity":9,"overall":7.98,"kind":"curated_estimate","confidence":"high","asOf":"2026-09-14","sources":["GTM AI pricing: https://gtm.ai/pricing","GTM AI MCP credits: https://gtm.ai/docs/mcp/credits","ZoomInfo: https://www.zoominfo.com"],"notes":"Strong enrichment peer when agents need MCP + verified contacts. PDL remains Depot pick for API-first $/record research pipelines; Apollo for GTM suites; ZoomInfo when quality+MCP matter more than unit price."},"lastReviewed":"2026-09-14","status":"active"},{"id":"enrich-layer","slug":"enrich-layer","name":"Enrich Layer (ex-Proxycurl)","website":"https://enrichlayer.com","tagline":"Budget enrichment peer for agents that outgrew Clearbit-free paths. Round 9 promoted high after pinning…","description":"Budget enrichment peer for agents that outgrew Clearbit-free paths. Round 9 promoted high after pinning annual Starter $588 / Pro $10,788 card. Prefer PDL/ZoomInfo/Apollo when verified GTM data + MCP matter more than raw credit cost. Strengths: Developer-classic LinkedIn-ish enrichment API; Cheap credits vs ZoomInfo list; Simple prepaid credits.","categoryIds":["cat-search-scrape"],"jtbdIds":["people-enrichment"],"regions":["global","us"],"pricingNotes":"As of 2026-09-14 (enrichlayer.com/pricing live): Annual credit subscriptions — Starter $588/yr (~17.5k contacts), Growth (~175k), Pro $10,788/yr (~629k), Ultra $22,788/yr (~1.4M). Monthly available; +2 months free on annual. Credits do not auto-overage; null results may still bill on some endpoints. Person/company/jobs/contact/logo APIs. PAYG top-ups also offered. Ex-Proxycurl rebrand.","strengths":["Developer-classic LinkedIn-ish enrichment API","Cheap credits vs ZoomInfo list","Simple prepaid credits"],"weaknesses":["Data provenance/ToS diligence required","Credit math varies by endpoint/params","Brand transition from Proxycurl"],"tags":["enrichment","proxycurl","people","company","api"],"scores":{"capabilityFit":8,"pricePerformance":8.6,"latencyReliability":7.8,"developerDx":8.4,"privacyPosture":5.5,"lockInRisk":3.5,"maturity":7.8,"overall":7.76,"kind":"curated_estimate","confidence":"high","asOf":"2026-09-14","sources":["Enrich Layer pricing: https://enrichlayer.com/pricing","How API costs are calculated: https://enrichlayer.com/docs/knowledge-base/billing/how-api-costs-are-calculated","Former Proxycurl users: https://enrichlayer.com/docs/knowledge-base/getting-started/former-proxycurl-user"],"notes":"Budget enrichment peer for agents that outgrew Clearbit-free paths. Round 9 promoted high after pinning annual Starter $588 / Pro $10,788 card. Prefer PDL/ZoomInfo/Apollo when verified GTM data + MCP matter more than raw credit cost."},"lastReviewed":"2026-09-14","status":"active"},{"id":"clearbit-hubspot","slug":"clearbit-hubspot","name":"Clearbit / HubSpot Enrichment","website":"https://www.hubspot.com","tagline":"Warning entry like Banana/PlayHT: tutorials saying Clearbit API → migrate to PDL/ZoomInfo/Enrich…","description":"Warning entry like Banana/PlayHT: tutorials saying Clearbit API → migrate to PDL/ZoomInfo/Enrich Layer/Apollo. Only stay if HubSpot is already the system of record. Strengths: Fine if you already live in HubSpot CRM; Standard enrichment bundled on paid seats; Intent/AI features on Credits.","categoryIds":["cat-search-scrape"],"jtbdIds":["people-enrichment"],"regions":["global","us"],"pricingNotes":"HISTORICAL / MIGRATED: Standalone Clearbit Enrichment API (person/company.clearbit.com) is sunset for new customers. HubSpot acquired Clearbit (2023); enrichment folded into HubSpot Smart CRM / Breeze. As of late 2025–2026: contact/company enrichment included on paid HubSpot seats (no HubSpot Credits for standard enrichment); Smart Properties / Buyer Intent still burn Credits (~$0.01/credit, $10/1k packs). Free HubSpot portals do not get enrichment. Do not build new agent pipelines on Clearbit API keys.","strengths":["Fine if you already live in HubSpot CRM","Standard enrichment bundled on paid seats","Intent/AI features on Credits"],"weaknesses":["Not an API-first enrichment product anymore","Requires HubSpot subscription","Legacy Clearbit endpoints dead for new builds"],"tags":["enrichment","sunset","hubspot","clearbit","warning"],"scores":{"capabilityFit":5.5,"pricePerformance":5,"latencyReliability":7,"developerDx":4,"privacyPosture":6.5,"lockInRisk":8,"maturity":4,"overall":5.12,"kind":"curated_estimate","confidence":"high","asOf":"2026-09-14","sources":["HubSpot Credits: https://www.hubspot.com/products/artificial-intelligence/credits","Clearbit by HubSpot help: https://help.clearbit.com/hc/en-us/articles/21570749175703-Clearbit-By-HubSpot-Set-Up-Clearbit-Enrichment-for-HubSpot","Clearbit pricing context 2026: https://www.landbase.com/blog/clearbit-pricing"],"notes":"Warning entry like Banana/PlayHT: tutorials saying Clearbit API → migrate to PDL/ZoomInfo/Enrich Layer/Apollo. Only stay if HubSpot is already the system of record."},"lastReviewed":"2026-09-14","status":"deprecated"},{"id":"unsloth","slug":"unsloth","name":"Unsloth","website":"https://unsloth.ai","tagline":"Depot OSS fine-tune default. Hosted LoRA → Predibase/Fireworks/Together; dead OpenPipe → do not resurrect…","description":"Depot OSS fine-tune default. Hosted LoRA → Predibase/Fireworks/Together; dead OpenPipe → do not resurrect. Run Unsloth on Modal/RunPod/HF Jobs. Strengths: Best-in-class OSS LoRA speed/VRAM efficiency; Huge community defaults; Works on rented GPUs.","categoryIds":["cat-train-finetune"],"jtbdIds":["finetune"],"regions":["global"],"pricingNotes":"As of 2026-09-14 (unsloth.ai/pricing): Free OSS local training (4/16-bit LoRA, Llama/Mistral/Gemma). Pro/Enterprise = contact-sales speed/VRAM multi-GPU claims (2.5×–30× marketing). No Unsloth-hosted GPU rate card — pair with HF Jobs / RunPod / Modal for cloud GPUs. HF blog documents free/cheap Unsloth+Hugging Face Jobs paths (T4-class ~$0.40/hr directional).","strengths":["Best-in-class OSS LoRA speed/VRAM efficiency","Huge community defaults","Works on rented GPUs"],"weaknesses":["Not a hosted fine-tune SaaS","Pro pricing opaque","You still buy GPUs elsewhere"],"tags":["finetune","lora","oss","training"],"scores":{"capabilityFit":8.8,"pricePerformance":9.2,"latencyReliability":7.5,"developerDx":9,"privacyPosture":8.5,"lockInRisk":2,"maturity":7.8,"overall":8.52,"kind":"curated_estimate","confidence":"high","asOf":"2026-09-14","sources":["Unsloth pricing: https://unsloth.ai/pricing","Unsloth: https://unsloth.ai","Unsloth + HF Jobs: https://huggingface.co/blog/unsloth-jobs"],"notes":"Depot OSS fine-tune default. Hosted LoRA → Predibase/Fireworks/Together; dead OpenPipe → do not resurrect. Run Unsloth on Modal/RunPod/HF Jobs."},"lastReviewed":"2026-09-14","status":"active"},{"id":"aws-sqs","slug":"aws-sqs","name":"Amazon SQS","website":"https://aws.amazon.com/sqs/","tagline":"Hyperscaler queue leaf when already on AWS. QStash remains Depot serverless HTTP bus default; SQS when…","description":"Hyperscaler queue leaf when already on AWS. QStash remains Depot serverless HTTP bus default; SQS when IAM/VPC gravity wins. Strengths: Ubiquitous AWS primitive; Cheap at modest volume + Free Tier; FIFO + DLQ patterns well understood.","categoryIds":["cat-cpu-compute"],"jtbdIds":["hyperscaler-agent-queue","serverless-agent-queue","event-driven-agent-jobs"],"regions":["global","us","eu"],"pricingNotes":"As of 2026-09-14 (aws.amazon.com/sqs/pricing): First 1M requests/mo free. Standard queues then billed per million requests (tiered; classic ~$0.40/M in many commercial regions — confirm current regional table). FIFO higher. 1–10 messages/request, 64KB chunks. No workflow state — pure queue.","strengths":["Ubiquitous AWS primitive","Cheap at modest volume + Free Tier","FIFO + DLQ patterns well understood"],"weaknesses":["Not HTTP-callback DX like QStash","You glue Lambdas/EventBridge yourself","Cross-region transfer charges"],"tags":["queue","aws","serverless","fifo"],"scores":{"capabilityFit":7.8,"pricePerformance":8.8,"latencyReliability":9.2,"developerDx":7,"privacyPosture":8,"lockInRisk":5.5,"maturity":9.5,"overall":8.03,"kind":"curated_estimate","confidence":"high","asOf":"2026-09-14","sources":["Amazon SQS pricing: https://aws.amazon.com/sqs/pricing/","Amazon SQS: https://aws.amazon.com/sqs/"],"notes":"Hyperscaler queue leaf when already on AWS. QStash remains Depot serverless HTTP bus default; SQS when IAM/VPC gravity wins."},"lastReviewed":"2026-09-14","status":"active"},{"id":"aws-eventbridge","slug":"aws-eventbridge","name":"Amazon EventBridge","website":"https://aws.amazon.com/eventbridge/","tagline":"AWS event backbone for agents (webhooks via API destinations, cron via Scheduler). Inngest/Temporal for…","description":"AWS event backbone for agents (webhooks via API destinations, cron via Scheduler). Inngest/Temporal for durable steps; QStash for HTTP-first without AWS. Strengths: Native event bus + Scheduler + Pipes + API destinations; Great AWS service fan-in for agents; Schema Registry free for custom/AWS schemas.","categoryIds":["cat-cpu-compute"],"jtbdIds":["hyperscaler-agent-queue","event-driven-agent-jobs"],"regions":["global","us","eu"],"pricingNotes":"As of 2026-09-14 (aws.amazon.com/eventbridge/pricing): Custom/partner bus events ~$1.00/M ingested; AWS management events ingest free; cross-account delivery often $1.00/M (custom) or $0.05/M (partner examples). Pipes ~$0.40/M requests after filter/batch. Scheduler: 14M invocations/mo free then ~$1.00/M. API destinations ~$0.20/M. 64KB chunk billing.","strengths":["Native event bus + Scheduler + Pipes + API destinations","Great AWS service fan-in for agents","Schema Registry free for custom/AWS schemas"],"weaknesses":["Cost surprise on custom event volume","AWS-centric","Not a multi-step workflow engine"],"tags":["events","aws","scheduler","pipes","bus"],"scores":{"capabilityFit":8.4,"pricePerformance":7.8,"latencyReliability":9.2,"developerDx":7.4,"privacyPosture":8,"lockInRisk":6,"maturity":9.4,"overall":8.01,"kind":"curated_estimate","confidence":"high","asOf":"2026-09-14","sources":["EventBridge pricing: https://aws.amazon.com/eventbridge/pricing/","Amazon EventBridge: https://aws.amazon.com/eventbridge/"],"notes":"AWS event backbone for agents (webhooks via API destinations, cron via Scheduler). Inngest/Temporal for durable steps; QStash for HTTP-first without AWS."},"lastReviewed":"2026-09-14","status":"active"},{"id":"growthbook","slug":"growthbook","name":"GrowthBook","website":"https://www.growthbook.io","tagline":"OSS/commercial peer to LaunchDarkly/Statsig for progressive agent rollout. GrowthBook when you want…","description":"OSS/commercial peer to LaunchDarkly/Statsig for progressive agent rollout. GrowthBook when you want warehouse-native + $40 seats; Unleash for flags-first self-host; LD when enterprise change-mgmt is law. Strengths: OSS + cheap Pro seats; Warehouse-native experimentation; Agent rollout + flags in one.","categoryIds":["cat-observability"],"jtbdIds":["progressive-agent-rollout","oss-feature-flags"],"regions":["global","us"],"pricingNotes":"As of 2026-09-14 (growthbook.io/pricing): Cloud Starter free (3 users, unlimited flags/experiments). Cloud Pro $40/seat/mo (up to 50 users) — CUPED, sequential, bandits, safe rollouts. Enterprise custom (SSO/SCIM/SLA). Self-host MIT core free (unlimited users). CDN overages $10/M requests, $1/GB; Managed Warehouse $0.03/1k events after 2M free. MCP server listed.","strengths":["OSS + cheap Pro seats","Warehouse-native experimentation","Agent rollout + flags in one"],"weaknesses":["Needs a warehouse for full value","Less flag-ops pedigree than LaunchDarkly","Enterprise still sales"],"tags":["feature-flags","experimentation","oss","rollout"],"scores":{"capabilityFit":8.6,"pricePerformance":9,"latencyReliability":8.2,"developerDx":8.8,"privacyPosture":8.2,"lockInRisk":3,"maturity":8,"overall":8.43,"kind":"curated_estimate","confidence":"high","asOf":"2026-09-14","sources":["GrowthBook pricing: https://www.growthbook.io/pricing","GrowthBook: https://www.growthbook.io"],"notes":"OSS/commercial peer to LaunchDarkly/Statsig for progressive agent rollout. GrowthBook when you want warehouse-native + $40 seats; Unleash for flags-first self-host; LD when enterprise change-mgmt is law."},"lastReviewed":"2026-09-14","status":"active"},{"id":"unleash","slug":"unleash","name":"Unleash","website":"https://www.getunleash.io","tagline":"Flags-first OSS peer. Prefer for kill-switches and audited rollouts; GrowthBook when experiments are the…","description":"Flags-first OSS peer. Prefer for kill-switches and audited rollouts; GrowthBook when experiments are the product; LaunchDarkly still Depot commercial default for large enterprises. Strengths: Flags-first with local SDK eval (no per-check bill); Strong OSS + EU residency options; Change-request approvals.","categoryIds":["cat-observability"],"jtbdIds":["progressive-agent-rollout","oss-feature-flags"],"regions":["global","us","eu"],"pricingNotes":"As of 2026-09-14 (getunleash.io/pricing): Open Source free self-host (limits: 1 project, 2 envs, 5k flags). Enterprise PAYG $75/seat/mo (5-seat min self-host), 53M API req/mo then $5/M, 99.9% uptime, card billing. Custom Enterprise annual (99.99%, CS, private instance/Edge add-ons, US/EU residency). No MAU metering.","strengths":["Flags-first with local SDK eval (no per-check bill)","Strong OSS + EU residency options","Change-request approvals"],"weaknesses":["Paid seats pricey vs GrowthBook","OSS edition misses SSO/RBAC depth","Light on experimentation vs GrowthBook"],"tags":["feature-flags","oss","rollout","eu"],"scores":{"capabilityFit":8.8,"pricePerformance":7.8,"latencyReliability":8.8,"developerDx":8.6,"privacyPosture":8.8,"lockInRisk":3.5,"maturity":8.8,"overall":8.43,"kind":"curated_estimate","confidence":"high","asOf":"2026-09-14","sources":["Unleash pricing: https://www.getunleash.io/pricing","Unleash: https://www.getunleash.io"],"notes":"Flags-first OSS peer. Prefer for kill-switches and audited rollouts; GrowthBook when experiments are the product; LaunchDarkly still Depot commercial default for large enterprises."},"lastReviewed":"2026-09-14","status":"active"},{"id":"confident-ai","slug":"confident-ai","name":"Confident AI","website":"https://www.confident-ai.com","tagline":"Best when your team already thinks in pytest/DeepEval. Braintrust remains Depot hosted-evals default…","description":"Best when your team already thinks in pytest/DeepEval. Braintrust remains Depot hosted-evals default; Promptfoo for OSS/CLI; Confident when DeepEval is the standard. Strengths: DeepEval pytest-native CI path; Shared datasets/metrics for eng+QA+PM; Cheap tracing ($1/GB-mo) vs peers.","categoryIds":["cat-observability"],"jtbdIds":["evals","oss-evals","llm-tracing","production-guardrail-evals"],"regions":["global","us"],"pricingNotes":"As of 2026-09-14 (confident-ai.com/pricing): Free forever (2 seats, 1 project, 5 test runs/week, 1 GB-mo traces). Starter $200/mo (5 projects, 5 GB-mo then $1/GB-mo). Team $2,000/mo (unlimited projects, 75 GB-mo). Enterprise custom (on-prem, HIPAA, residency). Online eval tokens ~$0.05/$0.40 per 1M in/out.","strengths":["DeepEval pytest-native CI path","Shared datasets/metrics for eng+QA+PM","Cheap tracing ($1/GB-mo) vs peers"],"weaknesses":["Starter jumps to $200/mo","Less brand gravity than Braintrust/LangSmith","Younger enterprise install base"],"tags":["evals","deepeval","ci","tracing","rag"],"scores":{"capabilityFit":8.6,"pricePerformance":7.8,"latencyReliability":8,"developerDx":9,"privacyPosture":7.8,"lockInRisk":4.5,"maturity":7.2,"overall":7.99,"kind":"curated_estimate","confidence":"high","asOf":"2026-09-14","sources":["Confident AI pricing: https://www.confident-ai.com/pricing","Confident AI: https://www.confident-ai.com","DeepEval: https://deepeval.com"],"notes":"Best when your team already thinks in pytest/DeepEval. Braintrust remains Depot hosted-evals default; Promptfoo for OSS/CLI; Confident when DeepEval is the standard."},"lastReviewed":"2026-09-14","status":"active"},{"id":"patronus","slug":"patronus","name":"Patronus AI","website":"https://www.patronus.ai","tagline":"Layer on when regulated hallucination/safety gates matter more than general CI evals. Lakera for injection…","description":"Layer on when regulated hallucination/safety gates matter more than general CI evals. Lakera for injection firewall; Galileo/Braintrust for broader prod evals. Strengths: Safety/hallucination-first evaluators; Regulated-industry narrative; Red-teaming / scenario testing.","categoryIds":["cat-observability"],"jtbdIds":["evals","production-guardrail-evals","prompt-injection-firewall","llm-input-guardrails"],"regions":["global","us"],"pricingNotes":"As of 2026-09-14: Enterprise evaluation/safety platform (Lynx hallucination detector, Glider judge). Public list price thin — secondary reports cite ~$25/mo Base + per-1k evaluator calls; treat as sales-led until official card pinned. Series-B backed.","strengths":["Safety/hallucination-first evaluators","Regulated-industry narrative","Red-teaming / scenario testing"],"weaknesses":["Opaque public pricing","Heavier than CI-first tools","Not a full tracing suite alone"],"tags":["evals","safety","hallucination","red-team"],"scores":{"capabilityFit":8.4,"pricePerformance":5.5,"latencyReliability":8,"developerDx":7.2,"privacyPosture":8.5,"lockInRisk":5.5,"maturity":7.5,"overall":7.29,"kind":"curated_estimate","confidence":"medium","asOf":"2026-09-14","sources":["Patronus AI: https://www.patronus.ai","Patronus overview 2026: https://atlan.com/know/ai-agent/ai-agent-evals/best-ai-agent-evaluation-platforms/"],"notes":"Layer on when regulated hallucination/safety gates matter more than general CI evals. Lakera for injection firewall; Galileo/Braintrust for broader prod evals."},"lastReviewed":"2026-09-14","status":"active"},{"id":"comet-opik","slug":"comet-opik","name":"Comet Opik","website":"https://www.comet.com/site/products/opik/","tagline":"Pick when you already live in Comet or want OSS Opik + optional cloud. Langfuse remains cheaper OSS default…","description":"Pick when you already live in Comet or want OSS Opik + optional cloud. Langfuse remains cheaper OSS default; Braintrust for eval-product teams. Strengths: OSS tracing+evals with Comet gravity; Good agent/RAG debug UX; Self-host escape hatch.","categoryIds":["cat-observability"],"jtbdIds":["evals","llm-tracing","oss-evals","experiment-tracking"],"regions":["global","us"],"pricingNotes":"As of 2026-09-14: Opik OSS self-host free. Comet Cloud free tier + Pro Cloud ~$19/mo class with ~100k spans / 60-day retention (confirm live comet.com/site/pricing). Fits teams already on Comet ML experiment tracking.","strengths":["OSS tracing+evals with Comet gravity","Good agent/RAG debug UX","Self-host escape hatch"],"weaknesses":["Comet packaging can confuse Opik-only buyers","Less eval-first than Braintrust/Confident","Cloud retention limits on cheap tiers"],"tags":["evals","tracing","oss","comet","rag"],"scores":{"capabilityFit":8.2,"pricePerformance":8.5,"latencyReliability":8,"developerDx":8.4,"privacyPosture":7.8,"lockInRisk":4,"maturity":8,"overall":8.04,"kind":"curated_estimate","confidence":"high","asOf":"2026-09-14","sources":["Comet Opik: https://www.comet.com/site/products/opik/","Opik GitHub: https://github.com/comet-ml/opik","Comet pricing: https://www.comet.com/site/pricing/"],"notes":"Pick when you already live in Comet or want OSS Opik + optional cloud. Langfuse remains cheaper OSS default; Braintrust for eval-product teams."},"lastReviewed":"2026-09-14","status":"active"},{"id":"langwatch","slug":"langwatch","name":"LangWatch","website":"https://langwatch.ai","tagline":"Strong when you need multi-turn/voice agent simulations with EU self-host. Langfuse default tracing…","description":"Strong when you need multi-turn/voice agent simulations with EU self-host. Langfuse default tracing; Braintrust default hosted evals. Strengths: Agent simulation + voice eval niche; EU-friendly + self-host; Transparent seat+event math.","categoryIds":["cat-observability"],"jtbdIds":["evals","llm-tracing","oss-evals","voice-agent-platform"],"regions":["eu","global"],"pricingNotes":"As of 2026-09-14 (langwatch.ai/pricing): Developer free (50k events/mo, 14-day access, 2 users). Growth €29/core-seat/mo + 200k events then €5/100k; 30-day retention then €3/GB. Enterprise hybrid/self-host/on-prem + marketplace billing. Scenario SDK OSS (Python+TS); voice agent simulation supported.","strengths":["Agent simulation + voice eval niche","EU-friendly + self-host","Transparent seat+event math"],"weaknesses":["Smaller ecosystem than Langfuse/LangSmith","Seat metering vs unlimited-seat rivals","EUR pricing FX for US buyers"],"tags":["evals","tracing","simulations","voice","eu","oss"],"scores":{"capabilityFit":8.4,"pricePerformance":8,"latencyReliability":8,"developerDx":8.6,"privacyPosture":8.5,"lockInRisk":3.5,"maturity":7.2,"overall":8.06,"kind":"curated_estimate","confidence":"high","asOf":"2026-09-14","sources":["LangWatch pricing: https://langwatch.ai/pricing","LangWatch: https://langwatch.ai"],"notes":"Strong when you need multi-turn/voice agent simulations with EU self-host. Langfuse default tracing; Braintrust default hosted evals."},"lastReviewed":"2026-09-14","status":"active"},{"id":"portkey","slug":"portkey","name":"Portkey","website":"https://portkey.ai","tagline":"Depot pick for production AI gateway (fallbacks, virtual keys, budgets). OpenRouter when you want a model…","description":"Depot pick for production AI gateway (fallbacks, virtual keys, budgets). OpenRouter when you want a model catalog/router; Helicone for request-path cost traces; Prisma AIRS when already in Palo Alto estate. Strengths: Universal API + fallbacks/load-balance/retries; Obs + prompt mgmt + guardrails in one; OSS self-host option.","categoryIds":["cat-observability","cat-inference-apis"],"jtbdIds":["llm-ai-gateway","multi-model-router","cost-monitoring","llm-tracing","llm-input-guardrails"],"regions":["global","us"],"pricingNotes":"As of 2026-09-14 (portkey.ai/pricing): Developer free forever (10k logs/mo, 3-day log / 30-day metrics retention). Production $49/mo (100k logs, +$9/100k overage, 30/90-day retention, guardrails, RBAC). Enterprise custom (10M+ logs, VPC, HIPAA/BAA, private cloud). OSS self-host gateway available. Note: packaging increasingly tied to Palo Alto Prisma AIRS AI Gateway narrative — same lineage as catalogued prisma-airs enterprise SKU.","strengths":["Universal API + fallbacks/load-balance/retries","Obs + prompt mgmt + guardrails in one","OSS self-host option"],"weaknesses":["Acqui/packaging drift into Prisma AIRS","Not a model marketplace like OpenRouter","Log caps on cheap tiers"],"tags":["gateway","routing","observability","guardrails","openai-compat"],"scores":{"capabilityFit":8.8,"pricePerformance":8.2,"latencyReliability":8.6,"developerDx":9,"privacyPosture":8.2,"lockInRisk":5,"maturity":8.2,"overall":8.31,"kind":"curated_estimate","confidence":"high","asOf":"2026-09-14","sources":["Portkey pricing: https://portkey.ai/pricing","Portkey: https://portkey.ai"],"notes":"Depot pick for production AI gateway (fallbacks, virtual keys, budgets). OpenRouter when you want a model catalog/router; Helicone for request-path cost traces; Prisma AIRS when already in Palo Alto estate."},"lastReviewed":"2026-09-14","status":"active"},{"id":"apify","slug":"apify","name":"Apify","website":"https://apify.com","tagline":"Depot pick when you need a scraper marketplace / reusable Actors, not just a crawl API. Firecrawl for…","description":"Depot pick when you need a scraper marketplace / reusable Actors, not just a crawl API. Firecrawl for LLM-ready extract; Bright Data for enterprise proxy; Scrapfly/ZenRows for simpler request APIs. Strengths: Huge Actor Store marketplace; Agent-payable prepaid tokens; Proxies + browsers + scheduling in one.","categoryIds":["cat-search-scrape","cat-sandboxes-browsers"],"jtbdIds":["scrape-actor-platform","web-scrape","serp-scrape","web-agent"],"regions":["global","eu","us"],"pricingNotes":"As of 2026-09-14 (apify.com/pricing): Free $0 ($5 usage). Starter $19/mo ($19 usage). Scale $199. Business $999. CU = 1 GB-RAM-hour: Free/Starter $0.20/CU, Scale $0.16, Business $0.13. Residential proxy $7–8/GB by plan. Agentic pay: x402/MPP prepaid tokens via agi.apify.com (no human account).","strengths":["Huge Actor Store marketplace","Agent-payable prepaid tokens","Proxies + browsers + scheduling in one"],"weaknesses":["CU math harder than per-request APIs","Failed runs still burn compute","Overkill for simple LLM-ready crawl"],"tags":["scrape","actors","proxy","marketplace","agents"],"scores":{"capabilityFit":9,"pricePerformance":7.5,"latencyReliability":8.6,"developerDx":8.4,"privacyPosture":7.5,"lockInRisk":4.5,"maturity":9.2,"overall":8.21,"kind":"curated_estimate","confidence":"high","asOf":"2026-09-14","sources":["Apify pricing: https://apify.com/pricing","Apify for agents: https://agi.apify.com/AGENTS.md"],"notes":"Depot pick when you need a scraper marketplace / reusable Actors, not just a crawl API. Firecrawl for LLM-ready extract; Bright Data for enterprise proxy; Scrapfly/ZenRows for simpler request APIs."},"lastReviewed":"2026-09-14","status":"active"},{"id":"scrapfly","slug":"scrapfly","name":"Scrapfly","website":"https://scrapfly.io","tagline":"Excellent Firecrawl peer when you need anti-bot + cloud browser with transparent credits. Prefer Firecrawl…","description":"Excellent Firecrawl peer when you need anti-bot + cloud browser with transparent credits. Prefer Firecrawl for simplest LLM markdown; Apify for Actor fleets. Strengths: Agent-readable pricing page; Unblocker + cloud browser + LLM extract; EU compliance posture.","categoryIds":["cat-search-scrape","cat-sandboxes-browsers"],"jtbdIds":["web-scrape","llm-ready-extract","enterprise-proxy"],"regions":["global","eu","us"],"pricingNotes":"As of 2026-09-14 (scrapfly.io/pricing machine-readable): Credit-based — Free 1k credits; Discovery $30/200k; Pro $100/1M (+$3.50/10k); Startup $250/2.5M; Enterprise $500/5.5M. Cloud browser (Playwright/CDP), Extraction API, Crawler, Screenshot share one balance. SOC2 + ISO27001; HIPAA on Custom.","strengths":["Agent-readable pricing page","Unblocker + cloud browser + LLM extract","EU compliance posture"],"weaknesses":["Credit multipliers hard to forecast","No published contractual uptime SLA on self-serve","Less Actor marketplace than Apify"],"tags":["scrape","proxy","browser","extraction","eu"],"scores":{"capabilityFit":8.6,"pricePerformance":8,"latencyReliability":8.2,"developerDx":8.8,"privacyPosture":8.5,"lockInRisk":3.5,"maturity":8,"overall":8.25,"kind":"curated_estimate","confidence":"high","asOf":"2026-09-14","sources":["Scrapfly pricing: https://scrapfly.io/pricing","Scrapfly: https://scrapfly.io"],"notes":"Excellent Firecrawl peer when you need anti-bot + cloud browser with transparent credits. Prefer Firecrawl for simplest LLM markdown; Apify for Actor fleets."},"lastReviewed":"2026-09-14","status":"active"},{"id":"zenrows","slug":"zenrows","name":"ZenRows","website":"https://www.zenrows.com","tagline":"Best per-request scrape API with agent MCP. Firecrawl when you want site-wide LLM markdown; Scrapfly when you…","description":"Best per-request scrape API with agent MCP. Firecrawl when you want site-wide LLM markdown; Scrapfly when you need cloud browser+extract suite. Strengths: Simple credit API for agents; MCP/Agent Toolkit first-class; Only successful requests billed.","categoryIds":["cat-search-scrape"],"jtbdIds":["web-scrape","llm-ready-extract","enterprise-proxy"],"regions":["global","us","eu"],"pricingNotes":"As of 2026-09-14 (zenrows.com/pricing): Free 5k credits/mo. Build $16/45k; Launch $57/250k; Growth $165/1.2M; Scale $456/5M; Enterprise custom. 1 credit = standard page; JS render 5; premium proxy 10; both 25; residential 25k credits/GB; Browser Sessions +5 credits/min. Agent Toolkit MCP/CLI. Annual ~17% off.","strengths":["Simple credit API for agents","MCP/Agent Toolkit first-class","Only successful requests billed"],"weaknesses":["Protected pages burn credits fast","Less crawl-site product than Firecrawl","No Actor marketplace"],"tags":["scrape","proxy","mcp","agents","api"],"scores":{"capabilityFit":8.4,"pricePerformance":8.5,"latencyReliability":8.4,"developerDx":8.8,"privacyPosture":7.8,"lockInRisk":3.5,"maturity":8.2,"overall":8.27,"kind":"curated_estimate","confidence":"high","asOf":"2026-09-14","sources":["ZenRows pricing: https://www.zenrows.com/pricing","ZenRows: https://www.zenrows.com"],"notes":"Best per-request scrape API with agent MCP. Firecrawl when you want site-wide LLM markdown; Scrapfly when you need cloud browser+extract suite."},"lastReviewed":"2026-09-14","status":"active"},{"id":"agentql","slug":"agentql","name":"AgentQL","website":"https://www.agentql.com","tagline":"Pinned Professional $99 on the live unauthenticated card (Round 8). Strong semantic query language for agent…","description":"Pinned Professional $99 on the live unauthenticated card (Round 8). Strong semantic query language for agent scrape/extract; use when CSS/XPath fragility dominates. Firecrawl remains LLM-crawl default; Apify for Actor marketplace. Strengths: Natural-language/structured query language for pages; MCP + Playwright SDKs; Resilient vs brittle CSS/XPath.","categoryIds":["cat-search-scrape","cat-sandboxes-browsers"],"jtbdIds":["llm-ready-extract","web-scrape","web-agent"],"regions":["global","us"],"pricingNotes":"As of 2026-09-14 (agentql.com/pricing live HTML): Free trial 300 API calls / 10 RPM / 1 hr remote browser / 1 concurrent. Starter $0/mo — 50 free API calls/mo then $0.02/call; 10 browser-hrs included then $0.12/hr; 5 concurrent. Professional $99/mo — 10,000 API calls included ($0.015 overage) + 500 browser-hrs ($0.10/hr) + 100 concurrent. Enterprise custom.","strengths":["Natural-language/structured query language for pages","MCP + Playwright SDKs","Resilient vs brittle CSS/XPath"],"weaknesses":["Paid rate card not fully pinned publicly","Smaller scale than Apify/Bright Data","Not a full proxy network"],"tags":["scrape","query-language","mcp","playwright","extract"],"scores":{"capabilityFit":8.2,"pricePerformance":7.8,"latencyReliability":7.6,"developerDx":8.8,"privacyPosture":7.2,"lockInRisk":4,"maturity":6.8,"overall":7.73,"kind":"curated_estimate","confidence":"high","asOf":"2026-09-14","sources":["AgentQL pricing: https://www.agentql.com/pricing","AgentQL MCP: https://docs.agentql.com/integrations/mcp","AgentQL: https://www.agentql.com"],"notes":"Pinned Professional $99 on the live unauthenticated card (Round 8). Strong semantic query language for agent scrape/extract; use when CSS/XPath fragility dominates. Firecrawl remains LLM-crawl default; Apify for Actor marketplace."},"lastReviewed":"2026-09-14","status":"active"},{"id":"verda","slug":"verda","name":"Verda","website":"https://verda.com","tagline":"Top EU list-price neocloud peer to Nebius. Prefer Nebius Depot default for global neocloud; Verda when EU…","description":"Top EU list-price neocloud peer to Nebius. Prefer Nebius Depot default for global neocloud; Verda when EU residency + public Blackwell card matter. Strengths: Full public on-demand+spot+serverless card; EU neocloud with IB clusters; Blackwell catalog live.","categoryIds":["cat-gpu-cloud","cat-cpu-compute"],"jtbdIds":["neocloud-gpu","predictable-gpu","batch-gpu","serverless-gpu","cheap-gpu"],"regions":["eu","global"],"pricingNotes":"As of 2026-10-27 (verda.com/pricing, formerly DataCrunch; fetched twice 2026-10-27, USD view; the page also has a € EUR view): On-demand 1x GPU instance per hour (stored rows): B200 $8.608, H200 $6.066, H100 SXM5 $4.652, A100 80GB $2.182, L40S $1.845; also listed, not stored: B300 $11.02, GB300 $12.67. Spot about half (e.g. H100 $2.326, H200 $3.033; not stored). Serverless containers priced separately (H100 $5.12/h continuous, $2.56 spot). Reserved 1 month to 2 years: 2% to 25% off on-demand. Instant IB clusters self-serve. SOC 2 + GDPR. On-demand USD prices have risen at every read since 2026-10-19; the page states no reason.","strengths":["Full public on-demand+spot+serverless card","EU neocloud with IB clusters","Blackwell catalog live"],"weaknesses":["On-demand H100 above Nebius/Voltage street","Less US hyperscale presence","Brand transition from DataCrunch"],"tags":["gpu","neocloud","eu","spot","serverless","infiniband"],"scores":{"capabilityFit":8.6,"pricePerformance":8.2,"latencyReliability":8.4,"developerDx":8.4,"privacyPosture":8.8,"lockInRisk":4,"maturity":7.8,"overall":8.23,"kind":"curated_estimate","confidence":"high","asOf":"2026-09-14","sources":["Verda pricing: https://verda.com/pricing","Verda H100: https://verda.com/h100"],"notes":"Top EU list-price neocloud peer to Nebius. Prefer Nebius Depot default for global neocloud; Verda when EU residency + public Blackwell card matter."},"lastReviewed":"2026-09-14","status":"active"},{"id":"thunder-compute","slug":"thunder-compute","name":"Thunder Compute","website":"https://www.thundercompute.com","tagline":"Indie-friendly A100/H100 when you want simpler than Vast marketplace. RunPod/Salad for interruptible fleets…","description":"Indie-friendly A100/H100 when you want simpler than Vast marketplace. RunPod/Salad for interruptible fleets; Nebius for SXM clusters. Strengths: Transparent single-GPU list prices; Per-minute + snapshots; No egress fees.","categoryIds":["cat-gpu-cloud"],"jtbdIds":["cheap-gpu","neocloud-gpu","batch-gpu"],"regions":["us","global"],"pricingNotes":"As of 2026-09-14 (thundercompute.com/pricing): Per-minute billing. A100 80GB $1.09/hr; H100 PCIe $3.20/hr; L40 $0.79; RTX A6000 $0.35. Multi-GPU listed (e.g. 2×A100 $2.18, 2×H100 $6.40 class). Snapshots $0.05/GB/mo; extra vCPU $0.04/hr; no egress fees claimed.","strengths":["Transparent single-GPU list prices","Per-minute + snapshots","No egress fees"],"weaknesses":["Narrower catalog than Verda/Nebius","H100 PCIe not SXM/IB training class","Younger brand"],"tags":["gpu","neocloud","cheap","a100","h100"],"scores":{"capabilityFit":7.2,"pricePerformance":8.8,"latencyReliability":7.4,"developerDx":8.2,"privacyPosture":7,"lockInRisk":3,"maturity":6.5,"overall":7.56,"kind":"curated_estimate","confidence":"high","asOf":"2026-09-14","sources":["Thunder Compute pricing: https://www.thundercompute.com/pricing","Thunder Compute: https://www.thundercompute.com"],"notes":"Indie-friendly A100/H100 when you want simpler than Vast marketplace. RunPod/Salad for interruptible fleets; Nebius for SXM clusters."},"lastReviewed":"2026-09-14","status":"active"},{"id":"hyperstack","slug":"hyperstack","name":"Hyperstack","website":"https://www.hyperstack.cloud","tagline":"Solid UK/EU GPU alternative with public card. Nebius/Verda first for neocloud; Hyperstack when UK entity +…","description":"Solid UK/EU GPU alternative with public card. Nebius/Verda first for neocloud; Hyperstack when UK entity + Studio inference matter. Strengths: Clear UK/EU GPU + inference Studio; Spot + reserved options; No bandwidth fees.","categoryIds":["cat-gpu-cloud","cat-inference-apis"],"jtbdIds":["neocloud-gpu","predictable-gpu","eu-sovereign-llm","cheap-gpu"],"regions":["eu","global"],"pricingNotes":"As of 2026-09-14 (hyperstack.cloud/gpu-pricing): On-demand H100 $2.50/hr, H100 NVLink $2.60, H100 SXM $3.20, H200 SXM $3.99, A100 $1.35–1.60, B200 $6.00, B300 $7.40, RTX Pro 6000 SE $1.85. Spot H100 PCIe from $2.00. Reserved H100 from ~$1.75. AI Studio tokens e.g. Llama 3.3 70B $0.80/$0.80; gpt-oss-120b $0.10/$0.40. No bandwidth fees claimed. UK-based.","strengths":["Clear UK/EU GPU + inference Studio","Spot + reserved options","No bandwidth fees"],"weaknesses":["Prepaid balance termination risk","Smaller than CoreWeave/Nebius","Support contact-gated for large deals"],"tags":["gpu","neocloud","eu","uk","inference"],"scores":{"capabilityFit":8.2,"pricePerformance":8.4,"latencyReliability":8,"developerDx":8,"privacyPosture":8.5,"lockInRisk":4,"maturity":7.5,"overall":7.98,"kind":"curated_estimate","confidence":"high","asOf":"2026-09-14","sources":["Hyperstack GPU pricing: https://www.hyperstack.cloud/gpu-pricing","Hyperstack: https://www.hyperstack.cloud"],"notes":"Solid UK/EU GPU alternative with public card. Nebius/Verda first for neocloud; Hyperstack when UK entity + Studio inference matter."},"lastReviewed":"2026-09-14","status":"active"},{"id":"gcore","slug":"gcore","name":"Gcore","website":"https://gcore.com","tagline":"EU inference bargain on published H100 €/hr. Pair with Mistral/Scaleway for model APIs; Verda when you need…","description":"EU inference bargain on published H100 €/hr. Pair with Mistral/Scaleway for model APIs; Verda when you need self-serve IB training clusters. Strengths: Aggressive inference H100 € list; EU digital sovereignty story; CDN/edge adjacency.","categoryIds":["cat-gpu-cloud","cat-cpu-compute"],"jtbdIds":["neocloud-gpu","eu-sovereign-llm","edge-deploy","predictable-gpu"],"regions":["eu","global"],"pricingNotes":"As of 2026-09-14 (gcore.com/pricing/ai): Everywhere Inference per-hour — 1×H100 €1.53, 1×A100 €1.26, 1×L40S €1.08 (ex-VAT). Virtual GPU 1×H100 €2.36/hr. Bare-metal 8×H100/H200/B300/GB300 clusters sales/calculator. Luxembourg-rooted edge/CDN company with EU sovereignty narrative.","strengths":["Aggressive inference H100 € list","EU digital sovereignty story","CDN/edge adjacency"],"weaknesses":["Bare-metal cluster list opaque vs Verda","VAT and multi-product complexity","Less AI-native DX than RunPod"],"tags":["gpu","eu","inference","edge","sovereign"],"scores":{"capabilityFit":8,"pricePerformance":8.8,"latencyReliability":8.2,"developerDx":7.2,"privacyPosture":8.8,"lockInRisk":4.5,"maturity":8.5,"overall":8.01,"kind":"curated_estimate","confidence":"high","asOf":"2026-09-14","sources":["Gcore AI pricing: https://gcore.com/pricing/ai","Gcore GPU Cloud: https://gcore.com/gpu-cloud"],"notes":"EU inference bargain on published H100 €/hr. Pair with Mistral/Scaleway for model APIs; Verda when you need self-serve IB training clusters."},"lastReviewed":"2026-09-14","status":"active"},{"id":"retell","slug":"retell","name":"Retell AI","website":"https://www.retellai.com","tagline":"Best Vapi peer for balanced inbound voice agents with clear $/min math. Vapi remains Depot default; Bland for…","description":"Best Vapi peer for balanced inbound voice agents with clear $/min math. Vapi remains Depot default; Bland for high-volume outbound; LiveKit for OSS media. Strengths: Transparent component rate card; No monthly platform fee; Strong inbound + builder DX.","categoryIds":["cat-comms","cat-media-gen"],"jtbdIds":["voice-agent-platform","voice-agents","speech-to-text","speech-synthesis"],"regions":["global","us"],"pricingNotes":"As of 2026-09-14 (retellai.com/pricing): True PAYG — platform voices all-in often ~$0.07–$0.31/min. Component: Voice Infra $0.055/min + TTS $0.015 (Retell/Cartesia/Minimax/Fish/OpenAI) or ElevenLabs $0.040 + LLM per-model/min + telephony ~$0.015/min US. 20 concurrent free then $8/concurrency/mo. HIPAA/BAA on Enterprise. No platform fee.","strengths":["Transparent component rate card","No monthly platform fee","Strong inbound + builder DX"],"weaknesses":["All-in still stacks with LLM+telco","Enterprise for HIPAA","Less outbound-dialer focus than Bland"],"tags":["voice","telephony","agents","stt","tts"],"scores":{"capabilityFit":8.8,"pricePerformance":7.8,"latencyReliability":8.4,"developerDx":9,"privacyPosture":8,"lockInRisk":4.5,"maturity":8,"overall":8.2,"kind":"curated_estimate","confidence":"high","asOf":"2026-09-14","sources":["Retell AI pricing: https://www.retellai.com/pricing","Retell AI: https://www.retellai.com"],"notes":"Best Vapi peer for balanced inbound voice agents with clear $/min math. Vapi remains Depot default; Bland for high-volume outbound; LiveKit for OSS media."},"lastReviewed":"2026-09-14","status":"active"},{"id":"bland","slug":"bland","name":"Bland AI","website":"https://www.bland.ai","tagline":"Pick for high-volume outbound when bundled $/min and dialer features beat BYOK flexibility. Vapi/Retell when…","description":"Pick for high-volume outbound when bundled $/min and dialer features beat BYOK flexibility. Vapi/Retell when you want provider choice; LiveKit when you own the stack. Strengths: All-in minute pricing (no model surprise); Outbound/campaign tooling; Regulated enterprise path.","categoryIds":["cat-comms","cat-media-gen"],"jtbdIds":["voice-agent-platform","voice-agents"],"regions":["global","us"],"pricingNotes":"As of 2026-09-14 (bland.ai/pricing): Bundled per-min (LLM+STT+TTS included, no token pass-through) — Start $0.14/min (10 concurrent, 100 calls/day, $0 platform); Build $0.12/min + $299/mo (50 concurrent, 2k calls/day); Enterprise custom (VPC/on-prem, BAA, SSO). Telephony separate / BYOT. 99.9% uptime listed.","strengths":["All-in minute pricing (no model surprise)","Outbound/campaign tooling","Regulated enterprise path"],"weaknesses":["Proprietary model (limited BYO LLM)","Platform fee on Build+","Compliance features Enterprise-gated"],"tags":["voice","telephony","outbound","agents","hipaa"],"scores":{"capabilityFit":8.6,"pricePerformance":7.5,"latencyReliability":8,"developerDx":8.2,"privacyPosture":8.2,"lockInRisk":6.5,"maturity":7.8,"overall":7.78,"kind":"curated_estimate","confidence":"high","asOf":"2026-09-14","sources":["Bland AI pricing: https://www.bland.ai/pricing","Bland AI: https://www.bland.ai"],"notes":"Pick for high-volume outbound when bundled $/min and dialer features beat BYOK flexibility. Vapi/Retell when you want provider choice; LiveKit when you own the stack."},"lastReviewed":"2026-09-14","status":"active"},{"id":"speechmatics","slug":"speechmatics","name":"Speechmatics","website":"https://www.speechmatics.com","tagline":"Depot pick for multilingual/production STT beyond English-first stacks. Deepgram remains default $/min…","description":"Depot pick for multilingual/production STT beyond English-first stacks. Deepgram remains default $/min English agent STT; Gladia for EU code-switch API peer. Strengths: 55+ languages + code-switching Melia; Strong accuracy reputation; On-prem/privacy deployments.","categoryIds":["cat-media-gen","cat-inference-apis"],"jtbdIds":["multilingual-stt","speech-to-text","production-stt","voice-agents"],"regions":["global","eu","us"],"pricingNotes":"As of 2026-09-14 (speechmatics.com/pricing): $100 free credit, no card. Pro PAYG (50 concurrent realtime, 10 file jobs/s); Enterprise volume/on-prem/custom. Enhanced vs Standard vs Melia-1 (multilingual code-switch) models. Volume auto-discount >500 hrs/mo/type; optional model-training opt-in −33% STT. SOC2/ISO27001/GDPR/HIPAA. Exact Enhanced/Standard $/hr confirm in portal (marketing shows plan structure more than full matrix).","strengths":["55+ languages + code-switching Melia","Strong accuracy reputation","On-prem/privacy deployments"],"weaknesses":["Full $/hr matrix less obvious than Deepgram","Training discount trades privacy","TTS still early (EN-first)"],"tags":["stt","multilingual","realtime","on-prem","eu"],"scores":{"capabilityFit":9,"pricePerformance":7.2,"latencyReliability":8.8,"developerDx":7.8,"privacyPosture":8.8,"lockInRisk":4,"maturity":9,"overall":8.24,"kind":"curated_estimate","confidence":"high","asOf":"2026-09-14","sources":["Speechmatics pricing: https://www.speechmatics.com/pricing","Speechmatics: https://www.speechmatics.com"],"notes":"Depot pick for multilingual/production STT beyond English-first stacks. Deepgram remains default $/min English agent STT; Gladia for EU code-switch API peer."},"lastReviewed":"2026-09-14","status":"active"},{"id":"gladia","slug":"gladia","name":"Gladia","website":"https://www.gladia.io","tagline":"EU multilingual STT with a now-pinned official $/hr card. Speechmatics remains Depot multilingual pick…","description":"EU multilingual STT with a now-pinned official $/hr card. Speechmatics remains Depot multilingual pick (breadth/on-prem); Deepgram English agent default; Gladia when you want EU residency + code-switch at published Starter rates. Strengths: EU-native multilingual STT; Code-switching + diarization; Agent/audio intelligence add-ons.","categoryIds":["cat-media-gen","cat-inference-apis"],"jtbdIds":["multilingual-stt","speech-to-text","production-stt"],"regions":["eu","global"],"pricingNotes":"As of 2026-09-14 (support.gladia.io help article as fetched then): Starter PAYG async $0.61/hr, realtime $0.75/hr (30 RT / 25 async concurrent). Growth commit from $0.20/hr async / $0.25/hr realtime (up to 67% off). Diarization, auto language detect, 100+ languages included. Credit-wallet billing (new accounts €50 grant; legacy 10 hr/mo still mentioned on product pages). GDPR/HIPAA/SOC2 Type 2. As of 2026-10-25 (https://www.gladia.io/pricing, HTTP 200 to plain curl, no login): the page shows EUR rates only: 'Starts at 0.16€ /hour' (async Self-serve, Solaria-3), 'Starts at 0.68€ /hour' (real-time Self-serve), 'Start with 50€ free credits'. The help article (page says 'Updated 23 Oct 2026') no longer lists rates: 'Rates, volume tiers, add-on prices and top-up bonuses are kept up to date on our pricing page.' The USD figures above are not re-priced or converted here; any change goes through the reviewed --catalog step.","strengths":["EU-native multilingual STT","Code-switching + diarization","Agent/audio intelligence add-ons"],"weaknesses":["Price card on https://www.gladia.io/pricing is in EUR only (fetched 2026-10-25, HTTP 200 to plain curl: 'Starts at 0.16€ /hour' async, 'Starts at 0.68€ /hour' real-time); no $/hr rate on the page, so Depot's USD figures stay as of 2026-09-14","Smaller than Deepgram/Speechmatics","Less telephony agent product"],"tags":["stt","multilingual","eu","diarization"],"scores":{"capabilityFit":8.2,"pricePerformance":7.8,"latencyReliability":8,"developerDx":8.2,"privacyPosture":8.8,"lockInRisk":3.5,"maturity":7.2,"overall":7.94,"kind":"curated_estimate","confidence":"high","asOf":"2026-09-14","sources":["Gladia transcription pricing (help center): https://support.gladia.io/article/understanding-our-transcription-pricing-pv1atikh8y9c8sw7sudm3rcy","Gladia pricing: https://www.gladia.io/pricing","Gladia changelog (credit billing): https://www.gladia.io/changelog"],"notes":"EU multilingual STT with a now-pinned official $/hr card. Speechmatics remains Depot multilingual pick (breadth/on-prem); Deepgram English agent default; Gladia when you want EU residency + code-switch at published Starter rates."},"lastReviewed":"2026-09-14","status":"active"},{"id":"memobase","slug":"memobase","name":"Memobase","website":"https://memobase.ai","tagline":"Depot pick for portable/user-profile memory (MCP across Claude/ChatGPT/agents). Mem0 remains general…","description":"Depot pick for portable/user-profile memory (MCP across Claude/ChatGPT/agents). Mem0 remains general hosted-agent-memory default; Zep for temporal graphs; Letta for MemGPT-style runtime. Strengths: MCP-native cross-assistant memory; Cheap Pro tier; Profile/event model + graph retrieval.","categoryIds":["cat-vector-memory"],"jtbdIds":["cross-session-user-memory","hosted-agent-memory","agent-memory-kv","rag-memory"],"regions":["global","us","cn"],"pricingNotes":"As of 2026-09-14 (memobase.ai): Free $0 / 500 credits/mo (store_memory=10 credits, search=1). Pro $9/mo / 5k credits. Unlimited $29/mo unlimited credits + SLA option. MCP at mcp.memobase.ai; APIs also via api.memobase.dev / docs.memobase.io. Profile+event memory architecture (user-centric, not flat vectors).","strengths":["MCP-native cross-assistant memory","Cheap Pro tier","Profile/event model + graph retrieval"],"weaknesses":["Product surface spans personal vault + developer API","Younger than Mem0","CN/global endpoint split to reason about"],"tags":["memory","mcp","profiles","oss","agents"],"scores":{"capabilityFit":8,"pricePerformance":9,"latencyReliability":7.4,"developerDx":8.6,"privacyPosture":8.2,"lockInRisk":4,"maturity":6.8,"overall":7.94,"kind":"curated_estimate","confidence":"high","asOf":"2026-09-14","sources":["Memobase: https://memobase.ai","Memobase docs: https://docs.memobase.io","memodb-io/memobase: https://github.com/memodb-io/memobase"],"notes":"Depot pick for portable/user-profile memory (MCP across Claude/ChatGPT/agents). Mem0 remains general hosted-agent-memory default; Zep for temporal graphs; Letta for MemGPT-style runtime."},"lastReviewed":"2026-09-14","status":"active"},{"id":"supermemory","slug":"supermemory","name":"Supermemory","website":"https://supermemory.ai","tagline":"Managed context engine when you want memory without running a vector DB. Mem0 for simpler fact API; Cognee…","description":"Managed context engine when you want memory without running a vector DB. Mem0 for simpler fact API; Cognee for graph-native OSS. Strengths: Clear credit + rate card; Sub-300ms recall claims; Connectors + self-host on Scale.","categoryIds":["cat-vector-memory"],"jtbdIds":["hosted-agent-memory","rag-memory","cross-session-user-memory"],"regions":["global","us"],"pricingNotes":"As of 2026-09-14 (supermemory.ai/pricing): Free $5 credits/mo; Pro $19 ($20 credits); Max $100 ($130); Scale $399 ($600) with SOC2/HIPAA/self-host; Enterprise custom. Rate card in SM tokens — Memory plain $0.005/1k (rich $0.010); SuperRAG text $0.001/1k; Search $0.005/1k queries; ops $0.10/1k. Dedupes unchanged content.","strengths":["Clear credit + rate card","Sub-300ms recall claims","Connectors + self-host on Scale"],"weaknesses":["Proprietary memory layer","Can get spendy at Scale","Less OSS than Mem0/Cognee"],"tags":["memory","rag","agents","hipaa"],"scores":{"capabilityFit":8.4,"pricePerformance":7.8,"latencyReliability":8,"developerDx":8.8,"privacyPosture":8.2,"lockInRisk":5.5,"maturity":7.2,"overall":7.88,"kind":"curated_estimate","confidence":"high","asOf":"2026-09-14","sources":["Supermemory pricing: https://supermemory.ai/pricing/","Supermemory: https://supermemory.ai"],"notes":"Managed context engine when you want memory without running a vector DB. Mem0 for simpler fact API; Cognee for graph-native OSS."},"lastReviewed":"2026-09-14","status":"active"},{"id":"cognee","slug":"cognee","name":"Cognee","website":"https://www.cognee.ai","tagline":"Graph memory when documents→ontology matter. Zep for temporal bipartite graphs; Mem0 for simplest hosted…","description":"Graph memory when documents→ontology matter. Zep for temporal bipartite graphs; Mem0 for simplest hosted memory; Letta for stateful runtime (ex-MemGPT). Strengths: Graph-native ECL memory; Generous free + cheap token meter; EU company + OSS self-host.","categoryIds":["cat-vector-memory"],"jtbdIds":["hosted-agent-memory","temporal-graph-memory","rag-memory","embedded-vector","knowledge-graph-rag"],"regions":["eu","global"],"pricingNotes":"As of 2026-09-14 (cognee.ai/pricing): Free $0 (1 workspace, 1M tokens). Standard $1.00/1M tokens + $5/extra workspace/mo. Enterprise BYOC engagement (6–24 mo packages, startup discount). Apache-2.0 OSS with SQLite/LanceDB/Kuzu defaults. ECL pipeline → knowledge graph.","strengths":["Graph-native ECL memory","Generous free + cheap token meter","EU company + OSS self-host"],"weaknesses":["Enterprise is services-heavy BYOC","Not a drop-in Mem0 API clone","Young commercial cloud"],"tags":["memory","knowledge-graph","oss","eu","rag"],"scores":{"capabilityFit":8.4,"pricePerformance":8.8,"latencyReliability":7.6,"developerDx":8.2,"privacyPosture":8.5,"lockInRisk":3,"maturity":7,"overall":8.09,"kind":"curated_estimate","confidence":"high","asOf":"2026-09-14","sources":["Cognee pricing: https://www.cognee.ai/pricing","Cognee: https://www.cognee.ai"],"notes":"Graph memory when documents→ontology matter. Zep for temporal bipartite graphs; Mem0 for simplest hosted memory; Letta for stateful runtime (ex-MemGPT)."},"lastReviewed":"2026-09-14","status":"active"},{"id":"azure-service-bus","slug":"azure-service-bus","name":"Azure Service Bus","website":"https://azure.microsoft.com/products/service-bus","tagline":"Azure twin to SQS/EventBridge for agent queues when already on Azure. EventBridge remains AWS hyperscaler…","description":"Azure twin to SQS/EventBridge for agent queues when already on Azure. EventBridge remains AWS hyperscaler pick; QStash for HTTP-first serverless. Strengths: Sessions/FIFO/transactions for agents; Native Azure Functions glue; Premium isolation for noisy neighbors.","categoryIds":["cat-cpu-compute"],"jtbdIds":["hyperscaler-agent-queue","serverless-agent-queue","event-driven-agent-jobs"],"regions":["global","us","eu"],"pricingNotes":"As of 2026-09-14 (azure.microsoft.com/pricing/details/service-bus): Tiered — Basic (ops-based, sessions/transactions limited); Standard (base fee + included ops then graduated $/M; topics/subscriptions); Premium (dedicated Messaging Units, predictable isolation). Exact regional $- masked on marketing — use Azure calculator. Strong sessions/FIFO/peek-lock/scheduled messages for agent work queues.","strengths":["Sessions/FIFO/transactions for agents","Native Azure Functions glue","Premium isolation for noisy neighbors"],"weaknesses":["Regional price opacity on public page","Azure-centric lock-in","Heavier than QStash HTTP DX"],"tags":["queue","azure","fifo","sessions","enterprise"],"scores":{"capabilityFit":8.6,"pricePerformance":7.5,"latencyReliability":9.2,"developerDx":7.2,"privacyPosture":8.5,"lockInRisk":6,"maturity":9.5,"overall":8.04,"kind":"curated_estimate","confidence":"high","asOf":"2026-09-14","sources":["Azure Service Bus pricing: https://azure.microsoft.com/en-us/pricing/details/service-bus/","Service Bus overview: https://learn.microsoft.com/en-us/azure/service-bus-messaging/service-bus-queues-topics-subscriptions"],"notes":"Azure twin to SQS/EventBridge for agent queues when already on Azure. EventBridge remains AWS hyperscaler pick; QStash for HTTP-first serverless."},"lastReviewed":"2026-09-14","status":"active"},{"id":"gcp-pubsub","slug":"gcp-pubsub","name":"Google Cloud Pub/Sub","website":"https://cloud.google.com/pubsub","tagline":"GCP hyperscaler bus for agent events. Prefer Service Bus on Azure for sessions; EventBridge on AWS; QStash…","description":"GCP hyperscaler bus for agent events. Prefer Service Bus on Azure for sessions; EventBridge on AWS; QStash when you want no cloud account gravity. Strengths: Simple global pub/sub fan-out; Generous free 10 GiB; Push to Cloud Run/Functions.","categoryIds":["cat-cpu-compute"],"jtbdIds":["hyperscaler-agent-queue","serverless-agent-queue","event-driven-agent-jobs"],"regions":["global","us","eu"],"pricingNotes":"As of 2026-09-14 (cloud.google.com/pubsub/pricing): Message ingestion/delivery throughput ~$40/TiB after first 10 GiB/mo free. Some export/import paths ~$50/TiB without that free tier. Ordering keys, push/pull, exactly-once (regional) for agent fan-out. Confirm current SKU table for Lite vs standard.","strengths":["Simple global pub/sub fan-out","Generous free 10 GiB","Push to Cloud Run/Functions"],"weaknesses":["Weaker FIFO/session semantics than Service Bus","GCP-centric","TiB math surprises vs per-request queues"],"tags":["queue","pubsub","gcp","fanout","serverless"],"scores":{"capabilityFit":8.2,"pricePerformance":8.2,"latencyReliability":9.2,"developerDx":7.6,"privacyPosture":8.2,"lockInRisk":5.5,"maturity":9.5,"overall":8.13,"kind":"curated_estimate","confidence":"high","asOf":"2026-09-14","sources":["Pub/Sub pricing: https://cloud.google.com/pubsub/pricing","Pub/Sub: https://cloud.google.com/pubsub"],"notes":"GCP hyperscaler bus for agent events. Prefer Service Bus on Azure for sessions; EventBridge on AWS; QStash when you want no cloud account gravity."},"lastReviewed":"2026-09-14","status":"active"},{"id":"voyage","slug":"voyage","name":"Voyage AI","website":"https://www.voyageai.com","tagline":"Depot pick for RAG rerank. Keep Cohere as general embeddings default (enterprise + multilingual maturity)…","description":"Depot pick for RAG rerank. Keep Cohere as general embeddings default (enterprise + multilingual maturity); switch to Voyage when retrieval nDCG dominates and you want the cheap lite SKU + rerank-3 in one vendor. Strengths: Retrieval-quality leader on many 2026 embed/rerank benches; Generous 200M free tokens + cheap lite SKUs; First-class rerank-3 alongside embeddings.","categoryIds":["cat-inference-apis","cat-vector-memory"],"jtbdIds":["embeddings","rag-rerank","rag-memory","batch-cheap-embeddings"],"regions":["global","us"],"pricingNotes":"As of 2026-09-14 (docs.voyageai.com/docs/pricing): voyage-4-lite $0.02/1M, voyage-4 $0.06, voyage-4-large / voyage-context-4 / voyage-code-4 $0.12 (200M free tokens on those families). rerank-3 $0.05/1M and rerank-3-lite $0.02/1M (200M free). Batch API 33% off, 12h window; Files storage $0.05/GB-mo. MongoDB-owned; also sold via Atlas automated embedding.","strengths":["Retrieval-quality leader on many 2026 embed/rerank benches","Generous 200M free tokens + cheap lite SKUs","First-class rerank-3 alongside embeddings"],"weaknesses":["Narrower ecosystem than OpenAI/Cohere embeddings","MongoDB packaging can confuse procurement","Fine-tuned dedicated instances are sales-led"],"tags":["embeddings","rerank","rag","mongodb"],"scores":{"capabilityFit":9.2,"pricePerformance":8.8,"latencyReliability":8.4,"developerDx":8.6,"privacyPosture":7.8,"lockInRisk":4.2,"maturity":7.8,"overall":8.4,"kind":"curated_estimate","confidence":"high","asOf":"2026-09-14","sources":["Voyage AI pricing: https://docs.voyageai.com/docs/pricing","Voyage AI: https://www.voyageai.com"],"notes":"Depot pick for RAG rerank. Keep Cohere as general embeddings default (enterprise + multilingual maturity); switch to Voyage when retrieval nDCG dominates and you want the cheap lite SKU + rerank-3 in one vendor."},"lastReviewed":"2026-09-14","status":"active"},{"id":"llama-parse","slug":"llama-parse","name":"LlamaParse","website":"https://www.llamaindex.ai","tagline":"Depot pick for document-parse (PDF/Office → LLM-ready markdown/JSON). Unstructured when you need ETL…","description":"Depot pick for document-parse (PDF/Office → LLM-ready markdown/JSON). Unstructured when you need ETL connectors/VPC pipelines; Reducto when agentic extraction accuracy on nasty forms beats LlamaParse in a bake-off. Strengths: Default parser in the LlamaIndex agent ecosystem; Tiered accuracy/cost (Fast→Agentic Plus) with auto mode; 130+ file types + extract/classify/split on the same credit meter.","categoryIds":["cat-search-scrape","cat-train-finetune"],"jtbdIds":["document-parse","llm-ready-extract","rag-memory"],"regions":["global","us","eu"],"pricingNotes":"As of 2026-09-14 (llamaindex.ai/pricing + developers.llamaindex.ai): 1,000 credits = $1.25 (NA/EU). Free 10k credits/mo; Starter $50/mo (40k credits); Pro 400k credits. v2 parse tiers: Fast 1 / Cost-effective 3 / Agentic 10 / Agentic Plus 45 credits/page. 48h parse cache is free to re-parse. SOC2/HIPAA/GDPR; VPC on AWS/Azure Marketplace.","strengths":["Default parser in the LlamaIndex agent ecosystem","Tiered accuracy/cost (Fast→Agentic Plus) with auto mode","130+ file types + extract/classify/split on the same credit meter"],"weaknesses":["Credit math is easy to blow on Agentic Plus (45 cr/page ≈ $0.056/page)","Not OSS (LlamaIndex framework is; LlamaParse is commercial)","Retention/cache defaults need a privacy pass"],"tags":["document-ai","pdf","rag","llamaindex","ocr"],"scores":{"capabilityFit":9,"pricePerformance":7.8,"latencyReliability":8.4,"developerDx":8.8,"privacyPosture":8,"lockInRisk":4.8,"maturity":8.2,"overall":8.22,"kind":"curated_estimate","confidence":"high","asOf":"2026-09-14","sources":["LlamaParse pricing: https://www.llamaindex.ai/pricing","LlamaParse credit rates: https://developers.llamaindex.ai/llamaparse/general/pricing/"],"notes":"Depot pick for document-parse (PDF/Office → LLM-ready markdown/JSON). Unstructured when you need ETL connectors/VPC pipelines; Reducto when agentic extraction accuracy on nasty forms beats LlamaParse in a bake-off."},"lastReviewed":"2026-09-14","status":"active"},{"id":"unstructured","slug":"unstructured","name":"Unstructured","website":"https://unstructured.io","tagline":"Best when you want connector-heavy document ETL or to self-host the OSS partitioner. LlamaParse remains the…","description":"Best when you want connector-heavy document ETL or to self-host the OSS partitioner. LlamaParse remains the agent-builder default parser; Reducto for max extraction accuracy. Strengths: Simple $0.015/page hosted card + generous OSS self-host; Broad source/destination connectors for RAG ETL; Partition strategies (Fast/Hi-Res/VLM) without credit gymnastics.","categoryIds":["cat-search-scrape","cat-train-finetune"],"jtbdIds":["document-parse","llm-ready-extract","rag-memory"],"regions":["global","us"],"pricingNotes":"As of 2026-09-14 (unstructured.io/pricing): 10,000 free pages (no card), then hosted API $0.015/page, all features included. Business = custom dedicated instance / VPC / multi-tenant. OSS library remains free to self-host. HIPAA / SOC2 Type 2 / GDPR / ISO 27001. 45+ file types + 40+ connectors.","strengths":["Simple $0.015/page hosted card + generous OSS self-host","Broad source/destination connectors for RAG ETL","Partition strategies (Fast/Hi-Res/VLM) without credit gymnastics"],"weaknesses":["Agentic form/table quality often trails LlamaParse/Reducto in bake-offs","Business SKU is sales-led","Hosted vs OSS feature split confuses first-time buyers"],"tags":["document-ai","etl","rag","oss","pdf"],"scores":{"capabilityFit":8.4,"pricePerformance":8.2,"latencyReliability":8.2,"developerDx":8,"privacyPosture":8.6,"lockInRisk":3.8,"maturity":8.2,"overall":8.12,"kind":"curated_estimate","confidence":"high","asOf":"2026-09-14","sources":["Unstructured pricing: https://unstructured.io/pricing","Unstructured: https://unstructured.io"],"notes":"Best when you want connector-heavy document ETL or to self-host the OSS partitioner. LlamaParse remains the agent-builder default parser; Reducto for max extraction accuracy."},"lastReviewed":"2026-09-14","status":"active"},{"id":"reducto","slug":"reducto","name":"Reducto","website":"https://reducto.ai","tagline":"Use when LlamaParse Agentic still drops tables/forms. Depot still picks LlamaParse for default document-parse…","description":"Use when LlamaParse Agentic still drops tables/forms. Depot still picks LlamaParse for default document-parse DX/ecosystem. Strengths: Strong agentic parse/extract on messy forms and tables; Clear per-1k-page endpoint card; EU/AU residency + ZDR on Growth.","categoryIds":["cat-search-scrape","cat-train-finetune"],"jtbdIds":["document-parse","llm-ready-extract","layout-aware-ocr"],"regions":["global","us","eu"],"pricingNotes":"As of 2026-09-14 (reducto.ai/pricing): Standard $150 free usage then list: r-1 Parse $10 / 1k pages, Extract $20, Deep Extract $40, Split $20, Classify $7.50, Edit $60 ($15 pre-filled). Growth = volume + ZDR/BAA + EU/AU residency. Enterprise VPC/on-prem. Up to $5k migration credits. 200 concurrent pages on Standard.","strengths":["Strong agentic parse/extract on messy forms and tables","Clear per-1k-page endpoint card","EU/AU residency + ZDR on Growth"],"weaknesses":["Extract/Edit SKUs get expensive vs LlamaParse Fast/Cost-effective","Younger ecosystem than LlamaIndex","Studio seat cap (5) on Standard"],"tags":["document-ai","pdf","extract","agentic-ocr"],"scores":{"capabilityFit":9,"pricePerformance":7.2,"latencyReliability":8.2,"developerDx":8.4,"privacyPosture":8.6,"lockInRisk":4.5,"maturity":7.4,"overall":8.02,"kind":"curated_estimate","confidence":"high","asOf":"2026-09-14","sources":["Reducto pricing: https://reducto.ai/pricing","Reducto: https://reducto.ai"],"notes":"Use when LlamaParse Agentic still drops tables/forms. Depot still picks LlamaParse for default document-parse DX/ecosystem."},"lastReviewed":"2026-09-14","status":"active"},{"id":"jina","slug":"jina","name":"Jina AI","website":"https://jina.ai","tagline":"Promoted high after pinning official models API token matrix ($0.05/1M). Best as EU-friendly all-in-one…","description":"Promoted high after pinning official models API token matrix ($0.05/1M). Best as EU-friendly all-in-one embed+rerank+Reader one-liner peer to Cohere/Voyage/Firecrawl. Depot embeddings default stays Cohere; rerank pick stays Voyage. Strengths: One vendor for embeddings + rerank + Reader (URL-to-markdown); r.jina.ai is a common one-liner in agent tools; EU HQ / multilingual embedding models.","categoryIds":["cat-inference-apis","cat-search-scrape"],"jtbdIds":["embeddings","rag-rerank","llm-ready-extract","web-scrape","agent-search","late-interaction-retrieval"],"regions":["global","eu"],"pricingNotes":"As of 2026-10-25 (GET api.jina.ai/v1/models): embeddings/rerank/reader-lm priced at pricing.prompt=0.00000005 → $0.05/1M tokens (jina-embeddings-v3/v4, jina-reranker-v3/v3.5, ReaderLM-v2, clip/colbert). jina-ocr-v1 $0.50/$2.00 per 1M prompt/completion. As of 2026-10-25 (https://jina.ai/embeddings/ FAQ): 'New users get an auto-generated API key with free tokens usable across any of our models.' The page states no free-token amount, so none is given here. Models API is source of record for prices. Same FAQ (2026-10-25): 'jina-embeddings-v4 is built on the Qwen2-VL base model, released under the Qwen Research License, which permits research and non-commercial use only. We therefore cannot license it commercially and provide it free of charge via the API instead.' and 'It is not suitable for production workloads'. The models API still lists jina-ai/jina-embeddings-v4 at pricing.prompt 0.00000005; Depot keeps the models-API figure (not re-priced here).","strengths":["One vendor for embeddings + rerank + Reader (URL-to-markdown)","r.jina.ai is a common one-liner in agent tools","EU HQ / multilingual embedding models"],"weaknesses":["No per-model price on https://jina.ai/embeddings/ (fetched 2026-10-25, HTTP 200: no $ amount anywhere in the page HTML); prices come only from GET api.jina.ai/v1/models","Reader is not a full anti-bot crawler (Firecrawl/ZenRows still win hostile sites)","Embed quality usually trails Voyage-4 / Cohere embed-v4 in RAG bake-offs"],"tags":["embeddings","rerank","reader","eu","rag"],"scores":{"capabilityFit":8.2,"pricePerformance":8.6,"latencyReliability":7.8,"developerDx":8.8,"privacyPosture":8.2,"lockInRisk":3.5,"maturity":7.6,"overall":8.12,"kind":"curated_estimate","confidence":"high","asOf":"2026-09-14","sources":["Jina models API (token matrix): https://api.jina.ai/v1/models","Jina Embeddings: https://jina.ai/embeddings/","Jina Reader: https://jina.ai/reader/","Jina AI: https://jina.ai"],"notes":"Promoted high after pinning official models API token matrix ($0.05/1M). Best as EU-friendly all-in-one embed+rerank+Reader one-liner peer to Cohere/Voyage/Firecrawl. Depot embeddings default stays Cohere; rerank pick stays Voyage."},"lastReviewed":"2026-09-14","status":"active"},{"id":"infomaniak","slug":"infomaniak","name":"Infomaniak AI","website":"https://www.infomaniak.com","tagline":"Fill the Swiss sovereign hole. Mistral remains Depot EU LLM default; Infomaniak when the RFP says Switzerland…","description":"Fill the Swiss sovereign hole. Mistral remains Depot EU LLM default; Infomaniak when the RFP says Switzerland + public open-model SKUs. Strengths: True Swiss/EU sovereign open-model API with a public CHF matrix; OpenAI-compatible — easy multi-home; Whisper + embeddings on the same bill.","categoryIds":["cat-inference-apis"],"jtbdIds":["eu-sovereign-llm","cheap-volume-llm","open-model-inference","speech-to-text"],"regions":["eu","ch"],"pricingNotes":"As of 2026-09-14 (infomaniak.com/en/hosting/ai-services/prices): Swiss-hosted OpenAI-compat API billed in CHF. Qwen3.5-122B-A10B-FP8 CHF 0.40 / 3.20 per 1M in/out; Ministral-3-14B-Instruct CHF 0.30 / 0.40; Whisper V3 CHF 0.006/min. Embeddings (Qwen3-Embedding-8B / bge-reranker) listed on the same card. 1M free credits for 1 month on signup. No training-on-your-data product.","strengths":["True Swiss/EU sovereign open-model API with a public CHF matrix","OpenAI-compatible — easy multi-home","Whisper + embeddings on the same bill"],"weaknesses":["Not frontier closed models (no Claude/GPT)","CHF billing + Swiss account KYC friction","Smaller catalog / rate-limit headroom than Mistral La Plateforme"],"tags":["eu","switzerland","sovereign","openai-compat","whisper"],"scores":{"capabilityFit":7.4,"pricePerformance":8,"latencyReliability":8.2,"developerDx":7.8,"privacyPosture":9.4,"lockInRisk":2.8,"maturity":8.4,"overall":7.97,"kind":"curated_estimate","confidence":"high","asOf":"2026-09-14","sources":["Infomaniak AI prices: https://www.infomaniak.com/en/hosting/ai-services/prices","Infomaniak AI services: https://www.infomaniak.com/en/hosting/ai-services"],"notes":"Fill the Swiss sovereign hole. Mistral remains Depot EU LLM default; Infomaniak when the RFP says Switzerland + public open-model SKUs."},"lastReviewed":"2026-09-14","status":"active"},{"id":"cloudferro","slug":"cloudferro","name":"CloudFerro","website":"https://cloudferro.com","tagline":"EU public-cloud GPU when Infomaniak/Mistral APIs are not enough and you must run weights in PL/DE…","description":"EU public-cloud GPU when Infomaniak/Mistral APIs are not enough and you must run weights in PL/DE. Nebius/Verda remain better $/GPU for commercial EU neocloud. Strengths: Public EU GPU list (Poland/Germany) with H100/H200/L40S; Sovereign/public-sector cloud credibility; Hourly through 1Y committed columns on the same table.","categoryIds":["cat-gpu-cloud","cat-cpu-compute"],"jtbdIds":["neocloud-gpu","cheap-gpu","predictable-gpu","eu-sovereign-llm"],"regions":["eu","pl","de"],"pricingNotes":"As of 2026-09-14 (cloudferro.com WAW4-1 GPU table, EUR ex-VAT): gpu.h100 on-demand €6.29/hr (1Y committed €2,776/mo); gpu.h200 €7.23/hr; shared L40S from €0.246/hr (1/8) to €1.970/hr (full). WAW/FRA OpenStack regions. Public Copernicus/EO heritage. H100 street is above Nebius/Verda — this is sovereignty + public list, not cheapest FLOPs.","strengths":["Public EU GPU list (Poland/Germany) with H100/H200/L40S","Sovereign/public-sector cloud credibility","Hourly through 1Y committed columns on the same table"],"weaknesses":["On-demand H100 ~€6.29/hr is expensive vs Nebius/Voltage","Not an inference API — you bring the serving stack","OpenStack DX vs hyperscaler/neocloud consoles"],"tags":["gpu","eu","poland","sovereign","h100"],"scores":{"capabilityFit":7.2,"pricePerformance":5.8,"latencyReliability":8.4,"developerDx":6.8,"privacyPosture":9.2,"lockInRisk":4,"maturity":8.6,"overall":7.32,"kind":"curated_estimate","confidence":"high","asOf":"2026-09-14","sources":["CloudFerro WAW4-1 GPU VMs: https://cloudferro.com/pricing/pricing-tables/waw4-1-cloud/virtual-machines-vm-with-gpu/","CloudFerro: https://cloudferro.com"],"notes":"EU public-cloud GPU when Infomaniak/Mistral APIs are not enough and you must run weights in PL/DE. Nebius/Verda remain better $/GPU for commercial EU neocloud."},"lastReviewed":"2026-09-14","status":"active"},{"id":"honeyhive","slug":"honeyhive","name":"HoneyHive","website":"https://www.honeyhive.ai","tagline":"Use the free Developer tier to stand up agent evals/traces in a day; Enterprise when you need self-host…","description":"Use the free Developer tier to stand up agent evals/traces in a day; Enterprise when you need self-host. Langfuse remains Depot tracing default; Braintrust hosted-evals default. Strengths: Real self-serve free eval+trace SKU (not just a waitlist); OTel-native + 50+ auto-instrumentations; Flexible enterprise hosting (self-host/hybrid).","categoryIds":["cat-observability"],"jtbdIds":["evals","llm-tracing","production-guardrail-evals"],"regions":["global","us"],"pricingNotes":"As of 2026-09-14 (honeyhive.ai/pricing): Developer free (no card) — 10k events/mo, 5 users, 1 workspace, 30-day retention, 1k RPM, evals + prompt versioning + CI. Enterprise custom (unlimited users/workspaces, SAML, custom retention, SLA, DPA/BAA, multi-tenant / single-tenant / hybrid / self-host). Data residency default AWS us-west-2. SOC2 Type II / GDPR / HIPAA. Startup discount < $5M raised.","strengths":["Real self-serve free eval+trace SKU (not just a waitlist)","OTel-native + 50+ auto-instrumentations","Flexible enterprise hosting (self-host/hybrid)"],"weaknesses":["No published mid-tier $ (free or sales)","10k events is small for production agents","Less brand gravity than Langfuse/Braintrust"],"tags":["evals","tracing","otel","agents","enterprise"],"scores":{"capabilityFit":8.4,"pricePerformance":7.8,"latencyReliability":8,"developerDx":8.6,"privacyPosture":8.8,"lockInRisk":4.2,"maturity":7.4,"overall":8.02,"kind":"curated_estimate","confidence":"high","asOf":"2026-09-14","sources":["HoneyHive pricing: https://www.honeyhive.ai/pricing","HoneyHive: https://www.honeyhive.ai"],"notes":"Use the free Developer tier to stand up agent evals/traces in a day; Enterprise when you need self-host. Langfuse remains Depot tracing default; Braintrust hosted-evals default."},"lastReviewed":"2026-09-14","status":"active"},{"id":"athina","slug":"athina","name":"Athina AI","website":"https://www.athina.ai","tagline":"Consider in a bake-off if HoneyHive's free tier is too small and you do not want Braintrust. Do not pick…","description":"Consider in a bake-off if HoneyHive's free tier is too small and you do not want Braintrust. Do not pick until a public SKU appears. Strengths: Team-friendly eval + monitoring UI; Datasets and production monitoring in one place; Fills the HoneyHive/Braintrust peer set.","categoryIds":["cat-observability"],"jtbdIds":["evals","llm-tracing","production-guardrail-evals"],"regions":["global"],"pricingNotes":"As of 2026-09-14: Collaborative LLM eval/observe platform. Official site is quote-led (no durable public $ table; /pricing 404). Secondary 2026 roundups cite a Team ~$299/mo and a free/low-log tier — treat as directional. Datasets, evals, monitoring, prompt management.","strengths":["Team-friendly eval + monitoring UI","Datasets and production monitoring in one place","Fills the HoneyHive/Braintrust peer set"],"weaknesses":["No official list price","Younger than Langfuse/Braintrust","Easy to overlap with tools you already pay for"],"tags":["evals","monitoring","prompts"],"scores":{"capabilityFit":8,"pricePerformance":6.8,"latencyReliability":7.6,"developerDx":8.2,"privacyPosture":7.4,"lockInRisk":4.8,"maturity":7,"overall":7.4,"kind":"curated_estimate","confidence":"medium","asOf":"2026-09-14","sources":["Athina AI: https://www.athina.ai"],"notes":"Consider in a bake-off if HoneyHive's free tier is too small and you do not want Braintrust. Do not pick until a public SKU appears."},"lastReviewed":"2026-09-14","status":"active"},{"id":"tigris","slug":"tigris","name":"Tigris","website":"https://www.tigrisdata.com","tagline":"R2 remains Depot object-storage / cheap-egress default (CDN+Workers gravity). Tigris when you want global…","description":"R2 remains Depot object-storage / cheap-egress default (CDN+Workers gravity). Tigris when you want global S3-compat without Cloudflare, or dynamic placement. Strengths: True $0 egress like R2, with a public request+storage card; Global dynamic placement (pay once); Agent-friendly (Fly-adjacent, S3 SDK).","categoryIds":["cat-object-storage"],"jtbdIds":["object-storage","cheap-egress-storage"],"regions":["global","us","eu"],"pricingNotes":"As of 2026-09-14 (tigrisdata.com/pricing.md): S3-compat, zero egress. Free tier 5 GB + 10k Class A / 100k Class B (no card). Global/single-region Standard $0.02/GB-mo; IA $0.01; Archive $0.004. Class A $0.005/1k, Class B $0.0005/1k. Multi-region Standard $0.025/GB-mo. console.storage.dev. Extraordinary bandwidth: email.","strengths":["True $0 egress like R2, with a public request+storage card","Global dynamic placement (pay once)","Agent-friendly (Fly-adjacent, S3 SDK)"],"weaknesses":["Smaller ecosystem than R2/S3","IA/Archive retrieval fees + min retention","Younger than Cloudflare/AWS"],"tags":["s3","egress-free","object-storage","global"],"scores":{"capabilityFit":8.2,"pricePerformance":9.2,"latencyReliability":8,"developerDx":8.6,"privacyPosture":7.8,"lockInRisk":3.2,"maturity":7,"overall":8.15,"kind":"curated_estimate","confidence":"high","asOf":"2026-09-14","sources":["Tigris pricing: https://www.tigrisdata.com/pricing.md","Tigris: https://www.tigrisdata.com"],"notes":"R2 remains Depot object-storage / cheap-egress default (CDN+Workers gravity). Tigris when you want global S3-compat without Cloudflare, or dynamic placement."},"lastReviewed":"2026-09-14","status":"active"},{"id":"flagsmith","slug":"flagsmith","name":"Flagsmith","website":"https://www.flagsmith.com","tagline":"GrowthBook remains Depot OSS flags pick (experimentation + flags). Flagsmith when you want flags-first…","description":"GrowthBook remains Depot OSS flags pick (experimentation + flags). Flagsmith when you want flags-first OpenFeature + a simpler cloud SKU than Unleash. Strengths: OSS + cheap cloud seats vs LaunchDarkly; OpenFeature-shaped flags for agent kill-switches; Honest request-meter with a usable free tier.","categoryIds":["cat-observability"],"jtbdIds":["oss-feature-flags","progressive-agent-rollout"],"regions":["global","eu"],"pricingNotes":"As of 2026-09-14 (flagsmith.com/pricing): Free 50k req/mo, 1 seat. Start-Up $45/mo ($40 annual) — 1M req, 3 seats, 2FA, A/B. Scale-Up $300/mo ($250 annual) — 5M+ req, 5 seats (extra $60/$50), SAML/SSO, change-requests. Overage ~$50/M then $7/100k on Start-Up after grace. OSS self-host; Enterprise cloud / private cloud / air-gap. OpenFeature-friendly.","strengths":["OSS + cheap cloud seats vs LaunchDarkly","OpenFeature-shaped flags for agent kill-switches","Honest request-meter with a usable free tier"],"weaknesses":["Seat counts get tight on Scale-Up","Less experimentation depth than Statsig/GrowthBook","Remote-eval latency depends on Edge/enterprise SKU"],"tags":["feature-flags","openfeature","oss","rollout"],"scores":{"capabilityFit":8,"pricePerformance":8.6,"latencyReliability":8.2,"developerDx":8.4,"privacyPosture":8.8,"lockInRisk":3,"maturity":8,"overall":8.21,"kind":"curated_estimate","confidence":"high","asOf":"2026-09-14","sources":["Flagsmith pricing: https://www.flagsmith.com/pricing","Flagsmith: https://www.flagsmith.com"],"notes":"GrowthBook remains Depot OSS flags pick (experimentation + flags). Flagsmith when you want flags-first OpenFeature + a simpler cloud SKU than Unleash."},"lastReviewed":"2026-09-14","status":"active"},{"id":"lightning-ai","slug":"lightning-ai","name":"Lightning AI","website":"https://lightning.ai","tagline":"Use when the team already lives in Lightning/PyTorch notebooks. Together/Unsloth remain Depot finetune…","description":"Use when the team already lives in Lightning/PyTorch notebooks. Together/Unsloth remain Depot finetune defaults; Modal for Python serverless agents. Strengths: Fastest path from notebook → multi-GPU train for Lightning/PyTorch users; Public GPU-hour card including H100; Studios + jobs + inference in one DX.","categoryIds":["cat-train-finetune","cat-gpu-cloud","cat-cpu-compute"],"jtbdIds":["finetune","serverless-gpu","ray-distributed-agents"],"regions":["global","us"],"pricingNotes":"As of 2026-09-14 (lightning.ai/pricing): Studio/cloud GPU hours on a public card — H100 New 80GB ~$4.50/hr on-demand (interruptible lower, ~$3.82–$4.58 class); T4/L4/A100 families also listed. Free Studio + credits to start. Team/seat SKUs (Pro/Teams) sit beside GPU-hour billing — confirm console. PyTorch Lightning / Fabric heritage.","strengths":["Fastest path from notebook → multi-GPU train for Lightning/PyTorch users","Public GPU-hour card including H100","Studios + jobs + inference in one DX"],"weaknesses":["H100 ~$4.50/hr is not a neocloud bargain vs Nebius/RunPod","Seat vs GPU-hour packaging is easy to double-pay","Less agent-sandbox DX than Modal"],"tags":["pytorch","studios","finetune","gpu","jobs"],"scores":{"capabilityFit":8.6,"pricePerformance":7,"latencyReliability":8.2,"developerDx":9,"privacyPosture":7.4,"lockInRisk":4.8,"maturity":8.4,"overall":7.93,"kind":"curated_estimate","confidence":"high","asOf":"2026-09-14","sources":["Lightning AI pricing: https://lightning.ai/pricing","Lightning AI: https://lightning.ai"],"notes":"Use when the team already lives in Lightning/PyTorch notebooks. Together/Unsloth remain Depot finetune defaults; Modal for Python serverless agents."},"lastReviewed":"2026-09-14","status":"active"},{"id":"black-forest-labs","slug":"black-forest-labs","name":"Black Forest Labs","website":"https://bfl.ai","tagline":"First-party FLUX when you want BFL terms/quality without a reseller markup. Fal remains Depot media-gen…","description":"First-party FLUX when you want BFL terms/quality without a reseller markup. Fal remains Depot media-gen default for multi-model catalogs. Strengths: Official FLUX.2 / FLUX 3 API (not only via Fal/Replicate resellers); Transparent credit = $0.01 card including video; EU lab with production FLUX weights.","categoryIds":["cat-media-gen"],"jtbdIds":["image-generation","video-generation","media-generation"],"regions":["global","eu","us"],"pricingNotes":"As of 2026-09-14 (docs.bfl.ai/quick_start/pricing): 1 credit = $0.01; API = Playground. FLUX.2 Klein 4B from $0.014/image, Klein 9B from $0.015, Pro from $0.03/MP (edit $0.045), Flex $0.05, Max $0.07. FLUX 3 video (t2v/i2v) $0.17/s hd / $0.29/s fhd, draft $0.06/s. FLUX.1 Kontext Pro $0.04/image. Fine-tuned Klein endpoints billed as base during public beta.","strengths":["Official FLUX.2 / FLUX 3 API (not only via Fal/Replicate resellers)","Transparent credit = $0.01 card including video","EU lab with production FLUX weights"],"weaknesses":["Single-family catalog (no Sora/Runway/Kling in-house)","Fal still wins if you need one key across many models","Video $/s adds up fast at FHD"],"tags":["flux","image","video","eu","diffusion"],"scores":{"capabilityFit":9,"pricePerformance":7.8,"latencyReliability":8.2,"developerDx":8.4,"privacyPosture":8,"lockInRisk":4,"maturity":7.8,"overall":8.14,"kind":"curated_estimate","confidence":"high","asOf":"2026-09-14","sources":["BFL API pricing: https://docs.bfl.ai/quick_start/pricing","Black Forest Labs: https://bfl.ai"],"notes":"First-party FLUX when you want BFL terms/quality without a reseller markup. Fal remains Depot media-gen default for multi-model catalogs."},"lastReviewed":"2026-09-14","status":"active"},{"id":"browserstack","slug":"browserstack","name":"BrowserStack","website":"https://www.browserstack.com","tagline":"Depot pick for mobile-computer-use on real phones. Cua if you only need Android VMs; AWS Device Farm if you…","description":"Depot pick for mobile-computer-use on real phones. Cua if you only need Android VMs; AWS Device Farm if you are already on AWS and want $0.17/min metered. Strengths: Largest self-serve real iOS/Android cloud for agents that must tap real devices; Appium/Espresso/XCUITest/Maestro/Flutter + REST; Public annual SKUs (not only RFP theater).","categoryIds":["cat-sandboxes-browsers"],"jtbdIds":["mobile-computer-use","cloud-browser","web-agent","visual-regression-qa"],"regions":["global"],"pricingNotes":"As of 2026-09-14 (browserstack.com/pricing): App Automate Device Cloud $199/mo billed annually (1 parallel, 30k+ real iOS/Android, unlimited minutes); Device Cloud Pro $249/mo. App Live Individual $39/mo; Team $150; Team Pro $249. Live Desktop $29 / Desktop+Mobile $39. Freelancer Live from $12.50/mo. Automate Desktop+Mobile from $175/mo per parallel. Enterprise add-ons (SSO, IP allowlist).","strengths":["Largest self-serve real iOS/Android cloud for agents that must tap real devices","Appium/Espresso/XCUITest/Maestro/Flutter + REST","Public annual SKUs (not only RFP theater)"],"weaknesses":["Priced for QA orgs, not indie agent sandboxes","Not a stealth computer-use browser (Anchor/Hyperbrowser still win the web CUA job)","Parallels get expensive fast"],"tags":["device-farm","android","ios","appium","real-devices"],"scores":{"capabilityFit":9,"pricePerformance":6.2,"latencyReliability":9,"developerDx":8,"privacyPosture":7.8,"lockInRisk":5.5,"maturity":9.4,"overall":7.95,"kind":"curated_estimate","confidence":"high","asOf":"2026-10-19","sources":["BrowserStack pricing: https://www.browserstack.com/pricing","BrowserStack: https://www.browserstack.com"],"notes":"Depot pick for mobile-computer-use on real phones. Cua if you only need Android VMs; AWS Device Farm if you are already on AWS and want $0.17/min metered."},"lastReviewed":"2026-10-19","status":"active"},{"id":"aws-device-farm","slug":"aws-device-farm","name":"AWS Device Farm","website":"https://aws.amazon.com/device-farm/","tagline":"AWS-estate default for mobile device farms. BrowserStack if you want the bigger self-serve catalog; Cua for…","description":"AWS-estate default for mobile device farms. BrowserStack if you want the bigger self-serve catalog; Cua for cheap Android VMs without real silicon. Strengths: IAM/VPC-native metered device minutes; Cheap desktop-browser instances at $0.005/min; Private device SKU for locked hardware.","categoryIds":["cat-sandboxes-browsers"],"jtbdIds":["mobile-computer-use"],"regions":["us"],"pricingNotes":"As of 2026-09-14 (aws.amazon.com/device-farm/pricing): Metered $0.17/device-minute after a one-time 1,000-minute free trial. Unmetered $250/slot/month (unlimited tests for that concurrency; Android/iOS and automated vs remote-access slots separate). Private dedicated devices from $200/mo. Desktop browser testing $0.005/instance-minute.","strengths":["IAM/VPC-native metered device minutes","Cheap desktop-browser instances at $0.005/min","Private device SKU for locked hardware"],"weaknesses":["Device catalog and DX trail BrowserStack","$0.17/min blows up long computer-use sessions (a 30-min loop is $5.10/device)","US-centric vs BrowserStack's 21 DCs"],"tags":["device-farm","android","ios","aws"],"scores":{"capabilityFit":7.8,"pricePerformance":7.4,"latencyReliability":8.6,"developerDx":7,"privacyPosture":8.8,"lockInRisk":5.8,"maturity":9,"overall":7.7,"kind":"curated_estimate","confidence":"high","asOf":"2026-09-14","sources":["AWS Device Farm pricing: https://aws.amazon.com/device-farm/pricing/","AWS Device Farm: https://aws.amazon.com/device-farm/"],"notes":"AWS-estate default for mobile device farms. BrowserStack if you want the bigger self-serve catalog; Cua for cheap Android VMs without real silicon."},"lastReviewed":"2026-09-14","status":"active"},{"id":"backblaze-b2","slug":"backblaze-b2","name":"Backblaze B2","website":"https://www.backblaze.com","tagline":"Best cheap always-hot archive for datasets/checkpoints. R2 still wins agent artifact serving with unbounded…","description":"Best cheap always-hot archive for datasets/checkpoints. R2 still wins agent artifact serving with unbounded free egress; Tigris for global $0 egress S3. Strengths: Dirt-cheap always-hot S3-compat; 3× free egress + partner zero-egress paths; Overdrive SKU aimed at training pipelines.","categoryIds":["cat-object-storage"],"jtbdIds":["object-storage","cheap-egress-storage"],"regions":["us","eu","global"],"pricingNotes":"As of 2026-09-14 (backblaze.com/cloud-storage/pricing): B2 $6.95/TB-mo (~$0.00695/GB), first 10 GB free, free API calls (A/B/C) on PAYG. Egress free up to 3× average monthly storage then $0.01/GB; unlimited free egress via CDN/compute partners (Cloudflare, Fastly, CoreWeave, …). B2 Overdrive from $15/TB-mo with unlimited egress (multi-PB commit) for AI/HPC.","strengths":["Dirt-cheap always-hot S3-compat","3× free egress + partner zero-egress paths","Overdrive SKU aimed at training pipelines"],"weaknesses":["Not $0 egress in the general case (R2/Tigris win that JTBD)","Fewer edge/compute hooks than R2+Workers","Overdrive is a serious commit"],"tags":["s3","b2","cheap-storage","ai-storage"],"scores":{"capabilityFit":8,"pricePerformance":9,"latencyReliability":8.8,"developerDx":8,"privacyPosture":8,"lockInRisk":3,"maturity":9.2,"overall":8.35,"kind":"curated_estimate","confidence":"high","asOf":"2026-09-14","sources":["Backblaze B2 pricing: https://www.backblaze.com/cloud-storage/pricing","Backblaze B2: https://www.backblaze.com/cloud-storage"],"notes":"Best cheap always-hot archive for datasets/checkpoints. R2 still wins agent artifact serving with unbounded free egress; Tigris for global $0 egress S3."},"lastReviewed":"2026-09-14","status":"active"},{"id":"moonshot","slug":"moonshot","name":"Moonshot (Kimi)","website":"https://www.moonshot.ai","tagline":"Long-context runner when Gemini procurement is blocked and you want 1M ctx at $3/$15. Google AI stays Depot…","description":"Long-context runner when Gemini procurement is blocked and you want 1M ctx at $3/$15. Google AI stays Depot long-context pick; Anthropic for coding-agent. Strengths: K3 1M context at a published $3/$15 — alternative to Gemini for long docs; K2.7 Code is a serious coding-agent SKU; OpenAI-compat international endpoint.","categoryIds":["cat-inference-apis"],"jtbdIds":["long-context-agents","coding-agent","cheap-volume-llm","open-model-inference"],"regions":["global","cn"],"pricingNotes":"As of 2026-09-14 (platform.kimi.ai/docs/pricing/chat; API api.moonshot.ai/v1 OpenAI-compat): kimi-k2.6 $0.95 in / $4 out per 1M (cached in $0.16), 262k ctx. kimi-k2.7-code $0.95/$4 (cached $0.19); highspeed $1.90/$8. kimi-k3 $3/$15 (cached $0.30), 1,048,576 ctx. File extract/storage APIs temporarily free.","strengths":["K3 1M context at a published $3/$15 — alternative to Gemini for long docs","K2.7 Code is a serious coding-agent SKU","OpenAI-compat international endpoint"],"weaknesses":["CN vendor / routing/compliance review for some customers","Tool-calling reliability still trails Claude for hard agents","Cache discount depends on hitting prefix cache"],"tags":["kimi","long-context","coding","china","openai-compat"],"scores":{"capabilityFit":8.6,"pricePerformance":8.4,"latencyReliability":7.8,"developerDx":8.2,"privacyPosture":6.2,"lockInRisk":4,"maturity":8,"overall":7.9,"kind":"curated_estimate","confidence":"high","asOf":"2026-09-14","sources":["Kimi chat pricing: https://platform.kimi.ai/docs/pricing/chat","Moonshot / Kimi platform: https://platform.kimi.ai"],"notes":"Long-context runner when Gemini procurement is blocked and you want 1M ctx at $3/$15. Google AI stays Depot long-context pick; Anthropic for coding-agent."},"lastReviewed":"2026-09-14","status":"active"},{"id":"parallel","slug":"parallel","name":"Parallel","website":"https://parallel.ai","tagline":"Strong Exa peer for agent search; wins when you need Task/deep-research in one bill. Exa remains Depot…","description":"Strong Exa peer for agent search; wins when you need Task/deep-research in one bill. Exa remains Depot agent-search default (neural search gravity). Strengths: Purpose-built agent web APIs (Search/Extract/Task/Monitor) with a full public matrix; Task API for deep research without stitching Exa+Firecrawl+LLM; Cheap turbo search at $0.001/request.","categoryIds":["cat-search-scrape"],"jtbdIds":["agent-search","search-augmented-llm","web-grounded-qa","people-enrichment","deep-research-agents"],"regions":["global","us"],"pricingNotes":"As of 2026-09-14 (docs.parallel.ai/getting-started/pricing): Search turbo/fast $1 / 1k requests (10 results); basic/advanced $5 / 1k; extra results $1 / 1k. Extract $1 / 1k URLs. Task processors $5–$2,400 / 1k successful runs (lite→ultra8x). Responses $10/$50/$250 per 1k by reasoning effort. Entity Search $5 / 1k (100 results). New orgs: $5 monthly credits (~5k turbo searches).","strengths":["Purpose-built agent web APIs (Search/Extract/Task/Monitor) with a full public matrix","Task API for deep research without stitching Exa+Firecrawl+LLM","Cheap turbo search at $0.001/request"],"weaknesses":["Ultra Task processors are expensive vs rolling your own","Younger than Exa/Tavily/Serper","Not a SERP-replica (keyword SEO tools still want Serper)"],"tags":["search","extract","deep-research","agents","web"],"scores":{"capabilityFit":8.8,"pricePerformance":8.4,"latencyReliability":8,"developerDx":8.8,"privacyPosture":7.4,"lockInRisk":4.2,"maturity":7.2,"overall":8.1,"kind":"curated_estimate","confidence":"high","asOf":"2026-09-14","sources":["Parallel API pricing: https://docs.parallel.ai/getting-started/pricing","Parallel: https://parallel.ai"],"notes":"Strong Exa peer for agent search; wins when you need Task/deep-research in one bill. Exa remains Depot agent-search default (neural search gravity)."},"lastReviewed":"2026-09-14","status":"active"},{"id":"logfire","slug":"logfire","name":"Pydantic Logfire","website":"https://pydantic.dev/logfire","tagline":"Best when the agent is already PydanticAI/Python and you want app+LLM traces together. Langfuse stays Depot…","description":"Best when the agent is already PydanticAI/Python and you want app+LLM traces together. Langfuse stays Depot tracing default; Braintrust for eval harnesses. Strengths: OTel traces for agents + app + infra on one $2/M meter; Honest free Personal cap (ingestion pauses — no surprise bill); First-class PydanticAI / Python DX.","categoryIds":["cat-observability"],"jtbdIds":["llm-tracing","evals","infra-observability"],"regions":["global","us","eu"],"pricingNotes":"As of 2026-09-14 (pydantic.dev/pricing): Personal free — 10M records/mo hard-capped, 1 seat, 3 projects, 30d. Team $49/mo (5 seats, $25 extra, 10M then $2/M, spend cap). Growth $249/mo unlimited seats/projects, up to 90d, BAA boilerplate. Enterprise cloud/dedicated/self-host. $0/1000 eval scores on Team+. AI Gateway BYOK (0% markup) / built-in 5% (3% on Growth).","strengths":["OTel traces for agents + app + infra on one $2/M meter","Honest free Personal cap (ingestion pauses — no surprise bill)","First-class PydanticAI / Python DX"],"weaknesses":["Not a dedicated LLM eval studio (Braintrust still deeper)","10M records vanishes if you log every token span","Gateway markup if you do not BYOK"],"tags":["otel","tracing","python","pydantic","evals"],"scores":{"capabilityFit":8.4,"pricePerformance":8.6,"latencyReliability":8.2,"developerDx":9.2,"privacyPosture":8.4,"lockInRisk":3.8,"maturity":8,"overall":8.33,"kind":"curated_estimate","confidence":"high","asOf":"2026-09-14","sources":["Pydantic Logfire pricing: https://pydantic.dev/pricing","Logfire: https://pydantic.dev/logfire"],"notes":"Best when the agent is already PydanticAI/Python and you want app+LLM traces together. Langfuse stays Depot tracing default; Braintrust for eval harnesses."},"lastReviewed":"2026-09-14","status":"active"},{"id":"minio","slug":"minio","name":"MinIO","website":"https://min.io","tagline":"Self-host object storage when R2/Tigris/B2 SaaS is not allowed. R2 remains managed cheap-egress default.","description":"Self-host object storage when R2/Tigris/B2 SaaS is not allowed. R2 remains managed cheap-egress default. Strengths: S3-compat you can run anywhere (the real zero-egress move); De-facto standard for on-prem/VPC buckets; No per-request tax once you own the disks.","categoryIds":["cat-object-storage"],"jtbdIds":["object-storage","cheap-egress-storage"],"regions":["global","self-host"],"pricingNotes":"As of 2026-09-14: OSS MinIO Community is free to self-host (S3-compat). Commercial AIStor / Enterprise is sales-led (support, SLA, AI features) — no durable public $/GB card. Use when you want object storage inside your VPC with zero vendor egress. Hardware and ops are your cost.","strengths":["S3-compat you can run anywhere (the real zero-egress move)","De-facto standard for on-prem/VPC buckets","No per-request tax once you own the disks"],"weaknesses":["You operate it (reliability is yours)","AIStor list price is opaque","Not a managed DX for indie agents (R2/Tigris are faster to first PUT)"],"tags":["s3","self-host","oss","object-storage","vpc"],"scores":{"capabilityFit":8.6,"pricePerformance":8.8,"latencyReliability":7.6,"developerDx":7.4,"privacyPosture":9.6,"lockInRisk":2.2,"maturity":9.2,"overall":8.41,"kind":"curated_estimate","confidence":"high","asOf":"2026-09-14","sources":["MinIO: https://min.io","MinIO AIStor: https://min.io/product/aistor"],"notes":"Self-host object storage when R2/Tigris/B2 SaaS is not allowed. R2 remains managed cheap-egress default."},"lastReviewed":"2026-09-14","status":"active"},{"id":"sauce-labs","slug":"sauce-labs","name":"Sauce Labs","website":"https://saucelabs.com","tagline":"BrowserStack peer for real-device mobile computer-use. Prefer when you already live in Sauce QA tooling or…","description":"BrowserStack peer for real-device mobile computer-use. Prefer when you already live in Sauce QA tooling or need the Real Device Access API. Depot mobile pick stays BrowserStack. Strengths: Mature real+virtual device clouds with Appium/Espresso/XCUITest; Real Device Access API for agent/programmable device control; Unlimited-minutes concurrency model (predictable vs per-minute).","categoryIds":["cat-sandboxes-browsers"],"jtbdIds":["mobile-computer-use","browser-agent-loop","cloud-browser"],"regions":["global","us","eu"],"pricingNotes":"As of 2026-09-14 (saucelabs.com/pricing): Live Testing $39/mo annual ($49 MoM). Virtual Device Cloud $149/mo annual ($199 MoM) — unlimited minutes, 1–24 parallel. Real Device Cloud $199/mo annual ($249 MoM) — real iOS/Android, Appium/Espresso/XCUITest, REST API. Real Device Access API (programmable device infra, GA 2026-02) is an add-on priced per private device, not per call. Enterprise: SSO, private devices, Sauce AI agents.","strengths":["Mature real+virtual device clouds with Appium/Espresso/XCUITest","Real Device Access API for agent/programmable device control","Unlimited-minutes concurrency model (predictable vs per-minute)"],"weaknesses":["Starts ~$149–$199/mo — heavier than indie BrowserStack freelancers","Real Device Access API requires Private Device add-on","QA-platform gravity; agent CUA docs thinner than BrowserStack/Cua"],"tags":["device-farm","mobile","appium","real-device","agents"],"scores":{"capabilityFit":8.8,"pricePerformance":6.2,"latencyReliability":8.8,"developerDx":7.8,"privacyPosture":8.2,"lockInRisk":4.8,"maturity":9.2,"overall":7.91,"kind":"curated_estimate","confidence":"high","asOf":"2026-09-14","sources":["Sauce Labs pricing: https://saucelabs.com/pricing","Sauce Labs Real Device Access API: https://siliconangle.com/2026/02/25/sauce-labs-launches-programmable-infrastructure-mobile-testing-real-device-access-api/","Sauce Labs: https://saucelabs.com"],"notes":"BrowserStack peer for real-device mobile computer-use. Prefer when you already live in Sauce QA tooling or need the Real Device Access API. Depot mobile pick stays BrowserStack."},"lastReviewed":"2026-09-14","status":"active"},{"id":"testmu-ai","slug":"testmu-ai","name":"TestMu AI (ex-LambdaTest)","website":"https://www.testmuai.com","tagline":"Budget BrowserStack/Sauce peer. Catalog as testmu-ai — do not add a second live LambdaTest SKU. Depot mobile…","description":"Budget BrowserStack/Sauce peer. Catalog as testmu-ai — do not add a second live LambdaTest SKU. Depot mobile pick stays BrowserStack unless price dominates. Strengths: Large real-device + browser matrix at aggressive list prices; Existing LambdaTest scripts/keys keep working post-rebrand; KaneAI agentic authoring for web/mobile.","categoryIds":["cat-sandboxes-browsers"],"jtbdIds":["mobile-computer-use","browser-agent-loop","cloud-browser"],"regions":["global","us","eu","in"],"pricingNotes":"As of 2026-09-14: LambdaTest rebranded to TestMu AI (Jan 2026) — same product/APIs/billing lineage (mobile-api.lambdatest.com still works). Free forever tier for limited mins. Live from ~$15/mo; web+mobile automation from ~$99/mo; Native App Automation historically ~$159/mo per parallel (annual). KaneAI/Kane CLI credit plans on testmuai.com/pricing (Starter $17 / Pro $89 / Max $179 annual). 10k+ real devices / 3k+ browsers. Verify live card — AI SKUs move fast.","strengths":["Large real-device + browser matrix at aggressive list prices","Existing LambdaTest scripts/keys keep working post-rebrand","KaneAI agentic authoring for web/mobile"],"weaknesses":["Brand/URL churn (testmuai.com vs lambdatest.com) confuses docs search","Cloud binary-upload latency same class as BrowserStack","AI credit SKUs layered on classic parallel pricing — quote carefully"],"tags":["device-farm","mobile","lambdatest","rebrand","kaneai"],"scores":{"capabilityFit":8.4,"pricePerformance":7.8,"latencyReliability":8.2,"developerDx":8,"privacyPosture":7.6,"lockInRisk":4.5,"maturity":8.6,"overall":7.94,"kind":"curated_estimate","confidence":"high","asOf":"2026-09-14","sources":["TestMu AI pricing: https://www.testmuai.com/pricing/","LambdaTest→TestMu AI Real Device Cloud: https://www.testmuai.com/blog/what-is-lambdatest-real-device-cloud/","TestMu AI: https://www.testmuai.com"],"notes":"Budget BrowserStack/Sauce peer. Catalog as testmu-ai — do not add a second live LambdaTest SKU. Depot mobile pick stays BrowserStack unless price dominates."},"lastReviewed":"2026-09-14","status":"active"},{"id":"genymotion-cloud","slug":"genymotion-cloud","name":"Genymotion Cloud","website":"https://www.genymotion.com","tagline":"Use when agents need Android VMs (not real handsets). Real-device mobile CUA stays BrowserStack/Sauce/AWS…","description":"Use when agents need Android VMs (not real handsets). Real-device mobile CUA stays BrowserStack/Sauce/AWS Device Farm; Cua covers multi-OS desktop+Android VM fleets. Strengths: True Android emulator fleets with simple $/min metering; CI/CD friendly; no physical-device queue contention; GPU/custom Premium for graphics-heavy agent apps.","categoryIds":["cat-sandboxes-browsers","cat-cpu-compute"],"jtbdIds":["mobile-computer-use","agent-sandbox"],"regions":["global","us","eu"],"pricingNotes":"As of 2026-09-14 (docs.genymotion.com/usage/saas/pricing): Pay-as-you-go $0.06 per running virtual-device minute (default 2 simultaneous; sales for more). Unlimited plan ~$219/device/mo or ~$179/device/mo annual (simultaneous devices). Premium = dedicated infra, custom CPU/RAM/GPU, SLA — quote. Android virtual devices in browser/CI — not physical handsets.","strengths":["True Android emulator fleets with simple $/min metering","CI/CD friendly; no physical-device queue contention","GPU/custom Premium for graphics-heavy agent apps"],"weaknesses":["Emulators ≠ real devices (carrier, OEM quirks, sensors)","iOS not in scope","Unlimited $179–$219/device adds up for wide matrices"],"tags":["android","emulator","device-farm","ci","agents"],"scores":{"capabilityFit":7.8,"pricePerformance":7.4,"latencyReliability":8.2,"developerDx":8,"privacyPosture":7.8,"lockInRisk":4,"maturity":8.4,"overall":7.75,"kind":"curated_estimate","confidence":"high","asOf":"2026-09-14","sources":["Genymotion SaaS pricing docs: https://docs.genymotion.com/usage/saas/pricing/","Genymotion Cloud: https://www.genymotion.com/product-cloud/","Genymotion pricing: https://www.genymotion.com/pricing/"],"notes":"Use when agents need Android VMs (not real handsets). Real-device mobile CUA stays BrowserStack/Sauce/AWS Device Farm; Cua covers multi-OS desktop+Android VM fleets."},"lastReviewed":"2026-09-14","status":"active"},{"id":"minimax","slug":"minimax","name":"MiniMax","website":"https://www.minimax.io","tagline":"Primary indie-usable Chinese multimodal API alongside Moonshot. Prefer Anthropic/OpenAI for coding agents…","description":"Primary indie-usable Chinese multimodal API alongside Moonshot. Prefer Anthropic/OpenAI for coding agents; MiniMax when you want one vendor for M3 + cheap TTS/video on intl endpoints. Strengths: International self-serve multimodal stack (text+TTS+video+image) with public token card; OpenAI- and Anthropic-compatible endpoints; Aggressive M3 promo pricing + long context.","categoryIds":["cat-inference-apis","cat-media-gen"],"jtbdIds":["frontier-chat","coding-agent","long-context-agents","speech-synthesis","video-generation","image-generation","multimodal","agent-tools"],"regions":["global","cn","us"],"pricingNotes":"As of 2026-09-14 (platform.minimax.io/docs/guides/pricing-paygo): International api.minimax.io (OpenAI + Anthropic-compat). MiniMax-M3 ≤512k: list $0.60/$2.40 → permanent 50% off $0.30/$1.20 per 1M (cache read $0.06); >512k $0.60/$2.40 after promo. Priority tier 1.5×. M2.7 $0.30/$1.20 (highspeed $0.60/$2.40). TTS speech-2.8-turbo $60/M chars, hd $100/M. ASR $0.38/hr. Video H3 768P $0.08/s / 2K $0.13/s. image-01 $0.0035. Token Plan (Plus/Max/Ultra) is closed to new purchases; its figure was removed. It is replaced by M Plan, listed at platform.minimax.io/docs/m-plan/monthly-offer as of 2026-10-21 at regular prices of Go $22/month, Explore $55/month and Build $132/month (first month 50% off; not stored as rows). Mainland keys use api.minimaxi.com — not interchangeable.","strengths":["International self-serve multimodal stack (text+TTS+video+image) with public token card","OpenAI- and Anthropic-compatible endpoints","Aggressive M3 promo pricing + long context"],"weaknesses":["CN vs intl platforms/keys are separate — easy to misconfigure","Music APIs discontinued for new users (2026-08-20)","Frontier quality still trails Anthropic/OpenAI/Gemini on hard agent benches"],"tags":["llm","multimodal","tts","video","china","international"],"scores":{"capabilityFit":8.4,"pricePerformance":8.8,"latencyReliability":7.8,"developerDx":8.2,"privacyPosture":6.2,"lockInRisk":4.8,"maturity":7.6,"overall":7.83,"kind":"curated_estimate","confidence":"high","asOf":"2026-09-14","sources":["MiniMax PAYG pricing: https://platform.minimax.io/docs/guides/pricing-paygo","MiniMax Token Plan: https://platform.minimax.io/docs/guides/pricing-token-plan","MiniMax platform: https://platform.minimax.io"],"notes":"Primary indie-usable Chinese multimodal API alongside Moonshot. Prefer Anthropic/OpenAI for coding agents; MiniMax when you want one vendor for M3 + cheap TTS/video on intl endpoints."},"lastReviewed":"2026-09-14","status":"active"},{"id":"dashscope","slug":"dashscope","name":"Alibaba Cloud Model Studio (DashScope)","website":"https://www.alibabacloud.com/help/en/model-studio/","tagline":"Catalog dashscope for Alibaba Model Studio / DashScope international Qwen. Complements Moonshot/MiniMax/Z.AI…","description":"Catalog dashscope for Alibaba Model Studio / DashScope international Qwen. Complements Moonshot/MiniMax/Z.AI for CN-origin models with indie-usable intl SKUs. Open-model hosts (Fireworks/Together) still win pure open-weight ops. Strengths: First-party Qwen with durable international public rate card; Multi-region Model Studio (SG/CN/EU/US/JP) + OpenAI-compat modes; Batch discounts and starter free quotas.","categoryIds":["cat-inference-apis","cat-media-gen"],"jtbdIds":["frontier-chat","coding-agent","long-context-agents","multimodal","cheap-volume-llm","embeddings"],"regions":["global","sg","cn","eu","us","jp"],"pricingNotes":"As of 2026-09-14 (alibabacloud.com/help/en/model-studio/model-pricing International): Qwen3.8-Max $2/$6 per 1M; Qwen3.7-Max $2.5/$7.5 (limited-time 50% off on some SKUs); Qwen3-Max $1.2/$6 (≤32k) tiering up at longer contexts; Qwen-Plus class ~$0.40/$1.20–$1.60. Batch often 50% off. Singapore free quotas (~1M tokens/90d) on many models. Intl endpoint dashscope-intl.aliyuncs.com; regional MaaS hosts for CN/EU/US/JP. Coding/Token Plans available in Singapore.","strengths":["First-party Qwen with durable international public rate card","Multi-region Model Studio (SG/CN/EU/US/JP) + OpenAI-compat modes","Batch discounts and starter free quotas"],"weaknesses":["Alibaba Cloud account/IAM friction vs pure API startups","Region/API-key non-interchangeability","Promo 50% offs expire — pin console before commits"],"tags":["qwen","alibaba","dashscope","llm","international"],"scores":{"capabilityFit":8.6,"pricePerformance":8.2,"latencyReliability":8.4,"developerDx":7.2,"privacyPosture":7,"lockInRisk":5.5,"maturity":8.8,"overall":7.86,"kind":"curated_estimate","confidence":"high","asOf":"2026-09-14","sources":["Model Studio model pricing: https://www.alibabacloud.com/help/en/model-studio/model-pricing","Model Studio overview: https://docs.modelstudio.console.alibabacloud.com/en/model-studio/what-is-model-studio","List models API: https://docs.modelstudio.console.alibabacloud.com/en/model-studio/list-models"],"notes":"Catalog dashscope for Alibaba Model Studio / DashScope international Qwen. Complements Moonshot/MiniMax/Z.AI for CN-origin models with indie-usable intl SKUs. Open-model hosts (Fireworks/Together) still win pure open-weight ops."},"lastReviewed":"2026-09-14","status":"active"},{"id":"zai","slug":"zai","name":"Z.AI (Zhipu GLM)","website":"https://z.ai","tagline":"International GLM via z.ai — catalog zai, not a second Zhipu mainland-only SKU. Strong cheap/coding runner…","description":"International GLM via z.ai — catalog zai, not a second Zhipu mainland-only SKU. Strong cheap/coding runner beside DeepSeek/Moonshot; Depot coding pick stays Anthropic. Strengths: Clear USD international token matrix for GLM-5.x family; Free Flash SKUs for prototyping; Vision/OCR/image/video/ASR in one catalog.","categoryIds":["cat-inference-apis","cat-media-gen"],"jtbdIds":["frontier-chat","coding-agent","budget-open-inference","cheap-volume-llm","multimodal","image-generation","speech-to-text"],"regions":["global","cn"],"pricingNotes":"As of 2026-09-14 (docs.z.ai/guides/overview/pricing): International OpenAI-compat api.z.ai. GLM-5.3 / 5.2 / 5.1 $1.40 in / $0.26 cached / $4.40 out per 1M; GLM-5 input $1, cached $0.20, output $3.20; GLM-4.7 input $0.60, cached $0.11, output $2.20; GLM-4.7-FlashX input $0.07, cached $0.01, output $0.40 (all per 1M, matching stored rows; re-checked on docs.z.ai/guides/overview/pricing 2026-10-21); GLM-4.7-Flash and GLM-4.5-Flash free. Vision GLM-4.6V $0.30/$0.90; GLM-OCR $0.03; GLM-Image $0.015/img; CogVideoX-3 $0.20/video; GLM-ASR-2512 $0.03/MTok (~$0.0024/min). Web Search tool $0.01/use. Cached-input storage limited-time free.","strengths":["Clear USD international token matrix for GLM-5.x family","Free Flash SKUs for prototyping","Vision/OCR/image/video/ASR in one catalog"],"weaknesses":["Brand split Z.AI vs Zhipu mainland consoles","Reliability/latency variance vs US frontier hosts","Enterprise DPA/ZDR story thinner than Anthropic/OpenAI"],"tags":["glm","zhipu","llm","china","international","flash"],"scores":{"capabilityFit":8.2,"pricePerformance":9,"latencyReliability":7.6,"developerDx":8,"privacyPosture":6,"lockInRisk":4.6,"maturity":8,"overall":7.79,"kind":"curated_estimate","confidence":"high","asOf":"2026-09-14","sources":["Z.AI pricing: https://docs.z.ai/guides/overview/pricing","Z.AI docs: https://docs.z.ai","Z.AI: https://z.ai"],"notes":"International GLM via z.ai — catalog zai, not a second Zhipu mainland-only SKU. Strong cheap/coding runner beside DeepSeek/Moonshot; Depot coding pick stays Anthropic."},"lastReviewed":"2026-09-14","status":"active"},{"id":"knock","slug":"knock","name":"Knock","website":"https://knock.app","tagline":"Best agent-native notification fabric (toolkit + human-in-the-loop). Courier remains Depot orchestration…","description":"Best agent-native notification fabric (toolkit + human-in-the-loop). Courier remains Depot orchestration default for broader maturity; prefer Knock when agents must trigger/approve notifications as tools. Strengths: First-class Agent Toolkit + AI agent workflow steps; Cross-channel prefs/batching/guides with clear free tier; HIPAA BAA on Enterprise.","categoryIds":["cat-comms"],"jtbdIds":["notification-orchestration","transactional-email","sms"],"regions":["global","us"],"pricingNotes":"As of 2026-09-14 (knock.app/pricing): Developer $0 — 10k messages, 500 guide active users, 500 AI agent credits (then $0.01/credit). Starter $250/mo — 50k messages then $0.005/msg, 2.5k guide users then $0.05, 2,000 AI agent credits. Enterprise custom (volume or monthly notified-user pricing, HIPAA/BAA, SAML/SCIM). Downstream ESP/SMS still billed separately. Agent Toolkit exposes workflows as agent tools; AI agent function credits by model+tokens (+3 min if web search).","strengths":["First-class Agent Toolkit + AI agent workflow steps","Cross-channel prefs/batching/guides with clear free tier","HIPAA BAA on Enterprise"],"weaknesses":["Starter jumps to $250/mo","Does not replace SendGrid/Twilio spend","Guide/active-user metering adds a second dimension"],"tags":["notifications","agents","email","push","orchestration"],"scores":{"capabilityFit":9,"pricePerformance":7.2,"latencyReliability":8.6,"developerDx":9,"privacyPosture":8.4,"lockInRisk":5,"maturity":8.4,"overall":8.22,"kind":"curated_estimate","confidence":"high","asOf":"2026-09-14","sources":["Knock pricing: https://knock.app/pricing","Knock Agent Toolkit: https://docs.knock.app/developer-tools/agent-toolkit/overview","Knock AI agent function: https://docs.knock.app/designing-workflows/ai-agent-function"],"notes":"Best agent-native notification fabric (toolkit + human-in-the-loop). Courier remains Depot orchestration default for broader maturity; prefer Knock when agents must trigger/approve notifications as tools."},"lastReviewed":"2026-09-14","status":"active"},{"id":"novu","slug":"novu","name":"Novu","website":"https://novu.co","tagline":"OSS-friendly Courier/Knock alternative. Pick when you want self-host or run-based economics; Knock for agent…","description":"OSS-friendly Courier/Knock alternative. Pick when you want self-host or run-based economics; Knock for agent toolkit; Courier for battle-tested multi-channel. Strengths: OSS + Cloud with agent-communication positioning; Run-based metering (not per-message surprises across channels); US+EU residency on all plans; self-host escape hatch.","categoryIds":["cat-comms"],"jtbdIds":["notification-orchestration"],"regions":["global","us","eu"],"pricingNotes":"As of 2026-09-14 (novu.co/pricing): Billed per workflow run (fan-out to N subscribers = N runs). Free $0 — 10k runs/mo, 20 workflows, US+EU residency, 3 seats. Pro from $30/mo — 30k+ runs, overage $1.20/1k runs. Team from $250/mo — 250k+ runs, RBAC, 10 envs. Enterprise custom (10M+, HIPAA BAA, SSO/SCIM, self-host). MIT OSS self-host free aside from infra.","strengths":["OSS + Cloud with agent-communication positioning","Run-based metering (not per-message surprises across channels)","US+EU residency on all plans; self-host escape hatch"],"weaknesses":["Topic fan-out can multiply run counts fast","Less agent-toolkit polish than Knock","Brand/docs still catching Courier/Knock gravity"],"tags":["notifications","oss","self-host","agents"],"scores":{"capabilityFit":8.2,"pricePerformance":8.6,"latencyReliability":8,"developerDx":8.4,"privacyPosture":8.8,"lockInRisk":2.8,"maturity":7.8,"overall":8.22,"kind":"curated_estimate","confidence":"high","asOf":"2026-09-14","sources":["Novu pricing: https://novu.co/pricing","Novu docs billing: https://docs.novu.co/platform/account/billing","Novu: https://novu.co"],"notes":"OSS-friendly Courier/Knock alternative. Pick when you want self-host or run-based economics; Knock for agent toolkit; Courier for battle-tested multi-channel."},"lastReviewed":"2026-09-14","status":"active"},{"id":"ably","slug":"ably","name":"Ably","website":"https://ably.com","tagline":"Depot pick for realtime agent↔user channels (token streaming, presence, multiplayer agent UIs). Not a…","description":"Depot pick for realtime agent↔user channels (token streaming, presence, multiplayer agent UIs). Not a substitute for email/SMS orchestration or durable workflow engines. Strengths: Battle-tested pub/sub + presence for streaming agent tokens to clients; Clear free→$29 path; strong SLAs on Enterprise; AI Transport scenarios documented on pricing page.","categoryIds":["cat-comms","cat-cpu-compute"],"jtbdIds":["realtime-agent-channel","notification-orchestration"],"regions":["global","us","eu"],"pricingNotes":"As of 2026-10-24 (ably.com/pricing, fetched 2026-10-24): Messages '$2.50 / million' ('As low as $0.50/M with volume discount'); channels and connections '$1.00 / million mins' each ('As low as $0.20/M with volume discount'); additional data transfer '$0.25 / GiB'. Packages: Free '$0' ('200 concurrent connections', '500 messages / second', '6M messages / month'); Standard '$29 / month + usage' ('10k concurrent channels', '10k concurrent connections'); Pro '$399 / month + usage' ('50k concurrent channels', '50k concurrent connections'); Enterprise 'Custom' ('99.999% uptime SLAs', 'HIPAA BAA', 'Dedicated cluster'). The page publishes no per-MAU price. Pricing examples on the page include an 'AI support chatbot'. See plan limits (/api/estimate) for the encoded plans.","strengths":["Battle-tested pub/sub + presence for streaming agent tokens to clients","Clear free→$29 path; strong SLAs on Enterprise","AI Transport scenarios documented on pricing page"],"weaknesses":["Usage math (msg + conn + channel mins) easy to under-estimate","Not a notification preference/orchestrator (pair with Knock/Courier)","Overkill for pure server-side agent queues (use QStash/SQS)"],"tags":["realtime","pubsub","websocket","agents","presence"],"scores":{"capabilityFit":8.8,"pricePerformance":7.4,"latencyReliability":9.2,"developerDx":8.4,"privacyPosture":8.6,"lockInRisk":5.2,"maturity":9.2,"overall":8.29,"kind":"curated_estimate","confidence":"high","asOf":"2026-09-14","sources":["Ably pricing: https://ably.com/pricing","Ably pub/sub pricing docs: https://ably.com/docs/platform/pricing","Ably: https://ably.com"],"notes":"Depot pick for realtime agent↔user channels (token streaming, presence, multiplayer agent UIs). Not a substitute for email/SMS orchestration or durable workflow engines."},"lastReviewed":"2026-09-14","status":"active"},{"id":"loops","slug":"loops","name":"Loops","website":"https://loops.so","tagline":"Product/lifecycle email for agent SaaS. Depot transactional default stays Resend; Postmark if deliverability…","description":"Product/lifecycle email for agent SaaS. Depot transactional default stays Resend; Postmark if deliverability crisis; Loops when marketing+txn share one audience DB. Strengths: Unified SaaS product email (lifecycle + transactional) with clean DX; Contact-based pricing — unlimited sends on paid; Workflows API / Claude connector for agent-assisted messaging.","categoryIds":["cat-comms"],"jtbdIds":["marketing-email","transactional-email"],"regions":["global","us"],"pricingNotes":"As of 2026-09-14 (loops.so/pricing + docs billing): Contact-based SaaS email (marketing + transactional). Free Early Stage — up to 1,000 subscribed contacts, ≤4,000 sends/30d, Loops branding. Paid tiers scale by subscribed contacts (transactional-only recipients don't count); unlimited sends on paid; no per-seat fee. Secondary trackers list paid entry ~$49/mo at 5k contacts — confirm live Stripe tier in console. Workflows API + Claude connector for agent-assisted campaigns.","strengths":["Unified SaaS product email (lifecycle + transactional) with clean DX","Contact-based pricing — unlimited sends on paid","Workflows API / Claude connector for agent-assisted messaging"],"weaknesses":["Not a deliverability specialist like Postmark for pure txn mail","Marketing contact growth drives bill even if send volume is flat","Free plan send cap + branding"],"tags":["email","saas","transactional","marketing","agents"],"scores":{"capabilityFit":8,"pricePerformance":7.6,"latencyReliability":8.2,"developerDx":8.8,"privacyPosture":7.8,"lockInRisk":4.2,"maturity":7.8,"overall":7.88,"kind":"curated_estimate","confidence":"high","asOf":"2026-09-14","sources":["Loops pricing: https://loops.so/pricing","Loops billing docs: https://loops.so/docs/account/billing","Loops: https://loops.so"],"notes":"Product/lifecycle email for agent SaaS. Depot transactional default stays Resend; Postmark if deliverability crisis; Loops when marketing+txn share one audience DB."},"lastReviewed":"2026-09-14","status":"active"},{"id":"cloudflare-ai-gateway","slug":"cloudflare-ai-gateway","name":"Cloudflare AI Gateway","website":"https://developers.cloudflare.com/ai-gateway/","tagline":"Free edge gateway peer to Portkey/OpenRouter. Depot llm-ai-gateway pick stays Portkey for virtual…","description":"Free edge gateway peer to Portkey/OpenRouter. Depot llm-ai-gateway pick stays Portkey for virtual keys/fallbacks/budgets; use CF AI Gateway when already on Workers and want $0 analytics+cache. Strengths: Free core multi-provider gateway on any CF account; Edge caching/rate limits + DLP/guardrails hooks; BYOK avoids the 5% Unified Billing tax.","categoryIds":["cat-observability","cat-inference-apis"],"jtbdIds":["llm-ai-gateway","cost-monitoring","llm-tracing","llm-rate-limit-shield"],"regions":["global","us","eu"],"pricingNotes":"As of 2026-09-14 (developers.cloudflare.com/ai-gateway/reference/pricing, updated 2026-05-19): Core gateway (analytics, caching, rate limiting, DLP profiles) free on all Cloudflare plans. Persistent logs: Workers Free 100k total; Workers Paid 10M/gateway. Unified Billing adds 5% fee on prepaid credits (provider token rates pass through, no markup). Guardrails billed as Workers AI tokens (Llama Guard). Logpush Workers Paid 10M req/mo then $0.05/M. Distinct catalog entry from cloudflare-workers-ai.","strengths":["Free core multi-provider gateway on any CF account","Edge caching/rate limits + DLP/guardrails hooks","BYOK avoids the 5% Unified Billing tax"],"weaknesses":["Deeper virtual-key/budget/guardrail UX still trails Portkey","Log retention tied to Workers plan limits","Unified Billing 5% surcharge if you buy credits through CF"],"tags":["gateway","cloudflare","caching","observability","byok"],"scores":{"capabilityFit":8,"pricePerformance":9.4,"latencyReliability":9,"developerDx":8.2,"privacyPosture":8.8,"lockInRisk":4,"maturity":8.4,"overall":8.41,"kind":"curated_estimate","confidence":"high","asOf":"2026-09-14","sources":["Cloudflare AI Gateway pricing: https://developers.cloudflare.com/ai-gateway/reference/pricing/","Cloudflare AI Gateway: https://developers.cloudflare.com/ai-gateway/","Unified Billing: https://developers.cloudflare.com/ai-gateway/features/unified-billing/"],"notes":"Free edge gateway peer to Portkey/OpenRouter. Depot llm-ai-gateway pick stays Portkey for virtual keys/fallbacks/budgets; use CF AI Gateway when already on Workers and want $0 analytics+cache."},"lastReviewed":"2026-09-14","status":"active"},{"id":"vercel-ai-gateway","slug":"vercel-ai-gateway","name":"Vercel AI Gateway","website":"https://vercel.com/ai-gateway","tagline":"Best when the app already ships on Vercel/AI SDK. Portkey remains production-gateway default; OpenRouter for…","description":"Best when the app already ships on Vercel/AI SDK. Portkey remains production-gateway default; OpenRouter for catalog breadth; CF AI Gateway for edge $0. Strengths: Zero-markup multi-provider gateway with AI SDK DX; Simple free credit on-ramp for prototypes; Model fallbacks/routing in one Vercel bill.","categoryIds":["cat-observability","cat-inference-apis"],"jtbdIds":["llm-ai-gateway","multi-model-router","cost-monitoring"],"regions":["global","us"],"pricingNotes":"As of 2026-09-14 (vercel.com/docs/ai-gateway/pricing): Zero platform fee / zero markup on provider list prices (incl. BYOK on paid). Free tier ~$5/mo credits on free-tier-eligible models with lower per-model rate limits; purchasing any AI Gateway Credits ends monthly free credit and unlocks full catalog + higher limits. BYOK requires paid credits path. Part of Vercel AI SDK gravity.","strengths":["Zero-markup multi-provider gateway with AI SDK DX","Simple free credit on-ramp for prototypes","Model fallbacks/routing in one Vercel bill"],"weaknesses":["Tied to Vercel account gravity","Free tier model subset + rate limits","Fewer enterprise virtual-key controls than Portkey"],"tags":["gateway","vercel","ai-sdk","router","byok"],"scores":{"capabilityFit":8.2,"pricePerformance":9,"latencyReliability":8.6,"developerDx":9.2,"privacyPosture":7.8,"lockInRisk":5.8,"maturity":8,"overall":8.21,"kind":"curated_estimate","confidence":"high","asOf":"2026-09-14","sources":["Vercel AI Gateway pricing: https://vercel.com/docs/ai-gateway/pricing","Vercel AI Gateway: https://vercel.com/ai-gateway","AI Gateway FAQ: https://vercel.com/docs/ai-gateway/faq"],"notes":"Best when the app already ships on Vercel/AI SDK. Portkey remains production-gateway default; OpenRouter for catalog breadth; CF AI Gateway for edge $0."},"lastReviewed":"2026-09-14","status":"active"},{"id":"turso","slug":"turso","name":"Turso","website":"https://turso.tech","tagline":"Depot pick for edge/per-tenant SQLite agent state. Serverless Postgres pick stays Neon; hot KV stays Redis…","description":"Depot pick for edge/per-tenant SQLite agent state. Serverless Postgres pick stays Neon; hot KV stays Redis; pgvector RAG stays Supabase. Strengths: Per-tenant/per-agent SQLite DBs are cheap and numerous; Edge replicas + embedded sync for local-first agents; Clear free→$5 path.","categoryIds":["cat-vector-memory","cat-object-storage"],"jtbdIds":["edge-sqlite-agent-db","agent-memory-kv","serverless-db"],"regions":["global","us","eu"],"pricingNotes":"As of 2026-09-14 (turso.tech/pricing): Free $0 — 100 DBs, 5GB storage, 500M rows read, 10M written, 3GB embedded sync, 1-day PITR. Developer $4.99/mo; Scaler $24.92/mo; Pro $416.58/mo (annual-save prices shown). Unlimited DBs on paid. Overage: storage/reads/writes/syncs per-unit. libSQL (SQLite-compatible) with edge replicas + embedded sync.","strengths":["Per-tenant/per-agent SQLite DBs are cheap and numerous","Edge replicas + embedded sync for local-first agents","Clear free→$5 path"],"weaknesses":["Rows-read metering can surprise chatty agents","Not a full Postgres (Neon/Supabase) or vector-first DB","Vector story is bring-your-own / extensions — not Pinecone-class"],"tags":["sqlite","libsql","edge","serverless","agents"],"scores":{"capabilityFit":7.8,"pricePerformance":9,"latencyReliability":8.4,"developerDx":8.8,"privacyPosture":8.2,"lockInRisk":3.6,"maturity":8,"overall":8.22,"kind":"curated_estimate","confidence":"high","asOf":"2026-09-14","sources":["Turso pricing: https://turso.tech/pricing","Turso unlimited databases: https://turso.tech/blog/turso-now-includes-unlimited-databases-for-everybody","Turso: https://turso.tech"],"notes":"Depot pick for edge/per-tenant SQLite agent state. Serverless Postgres pick stays Neon; hot KV stays Redis; pgvector RAG stays Supabase."},"lastReviewed":"2026-09-14","status":"active"},{"id":"hatchet","slug":"hatchet","name":"Hatchet","website":"https://hatchet.run","tagline":"Temporal/Inngest peer for agent task orchestration. Prefer Temporal for serious durability; Inngest for TS…","description":"Temporal/Inngest peer for agent task orchestration. Prefer Temporal for serious durability; Inngest for TS DX; Hatchet when you want OSS+simple run pricing. Strengths: OSS durable task engine aimed at AI agent workloads; Simple run-based cloud metering with free 100k; Self-host / BYOC escape hatch.","categoryIds":["cat-cpu-compute","cat-sandboxes-browsers"],"jtbdIds":["durable-agent-workflows","event-driven-agent-jobs","ray-distributed-agents"],"regions":["global","us"],"pricingNotes":"As of 2026-09-14 (hatchet.run/pricing): Developer Free + usage — first 100k task runs included then $10 per 1M runs; SOC2. Team $500/mo + usage (10 users, 5 tenants, 3-day retention, 500 RPS). Scale $1,000/mo + usage (unlimited users/tenants, 7-day retention, audit/HIPAA). Enterprise custom (SSO, BYOC, SLAs). Apache-2.0 self-host free.","strengths":["OSS durable task engine aimed at AI agent workloads","Simple run-based cloud metering with free 100k","Self-host / BYOC escape hatch"],"weaknesses":["Team tier starts at $500/mo — steep vs Inngest DX free tiers","Younger ecosystem than Temporal","Retention windows short on mid tiers"],"tags":["durable-execution","workflows","agents","oss","tasks"],"scores":{"capabilityFit":8.2,"pricePerformance":7.8,"latencyReliability":8.4,"developerDx":8.6,"privacyPosture":8.2,"lockInRisk":3.2,"maturity":7.2,"overall":8.02,"kind":"curated_estimate","confidence":"high","asOf":"2026-09-14","sources":["Hatchet Cloud pricing: https://hatchet.run/pricing","Hatchet durable tasks docs: https://docs.hatchet.run/v1/durable-tasks","Hatchet: https://hatchet.run"],"notes":"Temporal/Inngest peer for agent task orchestration. Prefer Temporal for serious durability; Inngest for TS DX; Hatchet when you want OSS+simple run pricing."},"lastReviewed":"2026-09-14","status":"active"},{"id":"dbos","slug":"dbos","name":"DBOS","website":"https://www.dbos.dev","tagline":"Embed durable steps in your app's Postgres rather than running a separate workflow cluster. Temporal remains…","description":"Embed durable steps in your app's Postgres rather than running a separate workflow cluster. Temporal remains Depot durability default. Strengths: Durable execution library with Postgres-backed exactly-once feel; Lightweight vs full Temporal clusters for app-embedded agents; Clear Pro/Teams checkpoint pricing.","categoryIds":["cat-cpu-compute"],"jtbdIds":["durable-agent-workflows","event-driven-agent-jobs"],"regions":["global","us"],"pricingNotes":"As of 2026-09-14 (dbos.dev/pricing): OSS DBOS Transact free (self-host). DBOS Pro $99/mo — 2 seats, manage apps, 1M free checkpoints/mo then $50/M. Teams $499/mo — 10 seats, 10 apps, 10M free checkpoints then $40/M, RBAC/alerting. Enterprise custom + self-hosted Conductor. Checkpoints = workflows/steps/transactions.","strengths":["Durable execution library with Postgres-backed exactly-once feel","Lightweight vs full Temporal clusters for app-embedded agents","Clear Pro/Teams checkpoint pricing"],"weaknesses":["Smaller ecosystem than Temporal/Inngest","Conductor tooling is the paid wedge","Less multi-language breadth historically"],"tags":["durable-execution","postgres","workflows","agents","oss"],"scores":{"capabilityFit":8,"pricePerformance":8,"latencyReliability":8.6,"developerDx":8.4,"privacyPosture":8.4,"lockInRisk":3.4,"maturity":7,"overall":7.99,"kind":"curated_estimate","confidence":"high","asOf":"2026-09-14","sources":["DBOS pricing: https://www.dbos.dev/pricing","DBOS: https://www.dbos.dev"],"notes":"Embed durable steps in your app's Postgres rather than running a separate workflow cluster. Temporal remains Depot durability default."},"lastReviewed":"2026-09-14","status":"active"},{"id":"restate","slug":"restate","name":"Restate","website":"https://restate.dev","tagline":"Strong Temporal alternative for durable AI agents, especially on Workers/Vercel. Remains medium until paid…","description":"Strong Temporal alternative for durable AI agents, especially on Workers/Vercel. Remains medium until paid $/action card is durably public. Depot durability pick stays Temporal Cloud. Strengths: Durable agents/workflows with journaled LLM/tool steps; Serverless invoker tunnel (Workers/Vercel/Deno); Free 50k actions + BYOC path.","categoryIds":["cat-cpu-compute","cat-sandboxes-browsers"],"jtbdIds":["durable-agent-workflows","event-driven-agent-jobs","stateful-agent-runtime","durable-idempotency"],"regions":["global","us","eu"],"pricingNotes":"As of 2026-09-14: Restate Cloud public usage-based — free tier 50k actions/mo, no card (restate.dev/blog/announcing-restate-cloud-public + docs). Paid action rates sales/console (marketing /pricing page thin to bots). BYOC (Jul 2026) for VPC-managed control plane. SOC2 / Enterprise SSO / HIPAA announced. Integrates invoker tunnel with Cloudflare Workers, Vercel Functions, Deno Deploy. ctx.get/set state reads/writes typically not billed as actions.","strengths":["Durable agents/workflows with journaled LLM/tool steps","Serverless invoker tunnel (Workers/Vercel/Deno)","Free 50k actions + BYOC path"],"weaknesses":["Public paid rate card still thin vs Temporal/Inngest","Younger mindshare","Action definition nuances need careful cost modeling"],"tags":["durable-execution","agents","workflows","serverless","byoc"],"scores":{"capabilityFit":8.6,"pricePerformance":7.8,"latencyReliability":8.8,"developerDx":8.4,"privacyPosture":8.4,"lockInRisk":3.8,"maturity":7.4,"overall":8.15,"kind":"curated_estimate","confidence":"medium","asOf":"2026-09-14","sources":["Restate Cloud public launch: https://restate.dev/blog/announcing-restate-cloud-public","Restate Cloud getting started: https://docs.restate.dev/cloud/getting-started","Restate pricing: https://restate.dev/pricing","Restate durable agents: https://docs.restate.dev/ai/patterns/durable-agents"],"notes":"Strong Temporal alternative for durable AI agents, especially on Workers/Vercel. Remains medium until paid $/action card is durably public. Depot durability pick stays Temporal Cloud."},"lastReviewed":"2026-09-14","status":"active"},{"id":"convex","slug":"convex","name":"Convex","website":"https://www.convex.dev","tagline":"Great TS agent product backend when you want reactive state+actions together. Prefer Supabase/Neon when…","description":"Great TS agent product backend when you want reactive state+actions together. Prefer Supabase/Neon when SQL/pgvector portability matters more than reactivity. Strengths: Reactive backend that fits agent UIs + tool actions in one system; Built-in vector search / crons / auth for agent products; Generous free tier for prototypes.","categoryIds":["cat-cpu-compute","cat-vector-memory"],"jtbdIds":["agent-backend","app-backend","serverless-db","agent-memory-kv"],"regions":["global","us"],"pricingNotes":"As of 2026-09-14 (convex.dev/pricing): Free & Starter $0 + PAYG — 1–6 developers, included function calls/bandwidth with overages (~$2.20/1M function calls, action compute $0.33/GB-hr bands). Professional $25/developer/mo. Business & Enterprise from $2,500/mo minimum (SAML/SSO, SLA). Reactive DB + vector search + crons + Node actions + auth.","strengths":["Reactive backend that fits agent UIs + tool actions in one system","Built-in vector search / crons / auth for agent products","Generous free tier for prototypes"],"weaknesses":["Professional per-developer pricing adds up","Not a drop-in Postgres; mental model shift","Business floor $2.5k/mo"],"tags":["backend","reactive","vector","agents","typescript"],"scores":{"capabilityFit":8.6,"pricePerformance":8,"latencyReliability":8.6,"developerDx":9.2,"privacyPosture":7.8,"lockInRisk":6.2,"maturity":8.4,"overall":8.15,"kind":"curated_estimate","confidence":"high","asOf":"2026-09-14","sources":["Convex pricing: https://www.convex.dev/pricing","Convex: https://www.convex.dev"],"notes":"Great TS agent product backend when you want reactive state+actions together. Prefer Supabase/Neon when SQL/pgvector portability matters more than reactivity."},"lastReviewed":"2026-09-14","status":"active"},{"id":"getstream","slug":"getstream","name":"Stream (GetStream)","website":"https://getstream.io","tagline":"Use when the agent product is a chat/feed UX. Ably remains Depot realtime channel pick for raw pub/sub token…","description":"Use when the agent product is a chat/feed UX. Ably remains Depot realtime channel pick for raw pub/sub token streaming; Stream when you need chat UX primitives. Strengths: Production chat/activity feeds with GenAI features for agent products; Strong SLA / enterprise story; SDKs cover web/mobile quickly.","categoryIds":["cat-comms"],"jtbdIds":["realtime-agent-channel","notification-orchestration"],"regions":["global","us","eu"],"pricingNotes":"As of 2026-09-14 (getstream.io/pricing): Chat Build free (makers higher limits; ~1k MAU / 100 concurrent class). Start $399/mo annual ($499 MoM) — 10k MAU. Elevate $599/mo annual — multi-tenancy, HIPAA, translations. Enterprise custom. Overage bands e.g. ~$0.09/MAU, ~$0.99/concurrent, ~$6/1M messages (confirm live). Video/Audio and Activity Feeds priced separately. AI moderation add-ons.","strengths":["Production chat/activity feeds with GenAI features for agent products","Strong SLA / enterprise story","SDKs cover web/mobile quickly"],"weaknesses":["Start tier ~$399/mo is steep vs Ably for pure token streaming","MAU+concurrent+message dimensions stack","Overkill if you only need server→client LLM token push"],"tags":["chat","realtime","feeds","agents","moderation"],"scores":{"capabilityFit":8.8,"pricePerformance":5.8,"latencyReliability":9,"developerDx":8.6,"privacyPosture":8.4,"lockInRisk":5.5,"maturity":9,"overall":7.94,"kind":"curated_estimate","confidence":"high","asOf":"2026-09-14","sources":["Stream Chat pricing: https://getstream.io/pricing/","Stream: https://getstream.io"],"notes":"Use when the agent product is a chat/feed UX. Ably remains Depot realtime channel pick for raw pub/sub token streaming; Stream when you need chat UX primitives."},"lastReviewed":"2026-09-14","status":"active"},{"id":"onesignal","slug":"onesignal","name":"OneSignal","website":"https://onesignal.com","tagline":"Push-first notify peer with pinned Free/$19 Growth card (Round 9 high). Prefer Knock/Courier when agents…","description":"Push-first notify peer with pinned Free/$19 Growth card (Round 9 high). Prefer Knock/Courier when agents orchestrate cross-channel workflows; OneSignal when mobile push MAU economics dominate. Strengths: Default mobile push infrastructure with huge free MAU band; Multi-channel (push/email/SMS/in-app) in one vendor; Mature SDKs and segmentation.","categoryIds":["cat-comms"],"jtbdIds":["notification-orchestration","sms","agent-push-notifications"],"regions":["global","us"],"pricingNotes":"As of 2026-09-14 (onesignal.com/pricing live HTML): Self-service Free $0/mo (all channels, basic automation/analytics, OneSignal AI & MCP). Growth starts at $19/mo (increased automation/integrations). Professional/Enterprise custom. Email: 20,000 free sends/mo then ~$1.50/1,000. Web push ~$0.004/subscriber; in-app messaging MAU bands (e.g. $0.012/MAU after free). SMS/RCS available as add-ons. Pin console before renewals — page still A/B-tests copy.","strengths":["Default mobile push infrastructure with huge free MAU band","Multi-channel (push/email/SMS/in-app) in one vendor","Mature SDKs and segmentation"],"weaknesses":["Weaker agent-tooling than Knock/Novu/Courier","Pricing page often experiment-heavy — pin before renewals","Email deliverability not Postmark-class"],"tags":["push","notifications","mobile","sms","email"],"scores":{"capabilityFit":8.4,"pricePerformance":8.2,"latencyReliability":8.6,"developerDx":8,"privacyPosture":7.6,"lockInRisk":4.8,"maturity":9,"overall":8.09,"kind":"curated_estimate","confidence":"high","asOf":"2026-09-14","sources":["OneSignal pricing: https://onesignal.com/pricing","OneSignal: https://onesignal.com"],"notes":"Push-first notify peer with pinned Free/$19 Growth card (Round 9 high). Prefer Knock/Courier when agents orchestrate cross-channel workflows; OneSignal when mobile push MAU economics dominate."},"lastReviewed":"2026-09-14","status":"active"},{"id":"chutes","slug":"chutes","name":"Chutes","website":"https://chutes.ai","tagline":"Budget open-model overflow with a now-pinned public TEE matrix. Depot open-model production pick stays…","description":"Budget open-model overflow with a now-pinned public TEE matrix. Depot open-model production pick stays Fireworks; ultra-budget volume DeepInfra/Novita; use Chutes as capacity valve and re-check live rates. Strengths: Aggressive open-model inference pricing; Serverless chute deploys for custom models; Useful overflow capacity for batch agents.","categoryIds":["cat-inference-apis","cat-gpu-cloud"],"jtbdIds":["budget-open-inference","cheap-volume-llm","open-model-inference","serverless-gpu","confidential-tee-inference"],"regions":["global","us"],"pricingNotes":"As of 2026-10-27 (chutes.ai/pricing, fetched twice 2026-10-27): Public TEE inference PAYG per 1M tokens, input/output (the page says per-model rates are 'pulled live from our pricing API'): gemma-4-31B-turbo $0.12/$0.37; Qwen3.8-27B $0.039/$1.25; Qwen3.6-27B $0.25/$2.00; DeepSeek-V4-Flash-0731 $0.44/$1.32; Kimi K2.6 $0.39/$1.80; Qwen 3.5 (397B-A17B) $0.45/$3.00; DeepSeek-V3.2 $1.00/$1.00; GLM-5.1 $0.90/$2.90; GLM-5.2 $1.25/$3.95; Kimi-K3 $1.85/$9.50 (cached-input rates also on page, not stored). Mistral-Nemo, Nemotron-3-Nano-Omni-30B, Qwen3-32B and Qwen3-235B-A22B-Thinking-2507 are no longer on the page (re-checked 2026-10-27; stored rows kept at their 2026-10-03 amounts, marked unverified, never ranked). Private Chute self-serve TEE GPU RTX Pro 6000 96 GB $1.80/hr + one-time deploy fee 3× hourly ($5.40). Page FAQ: 'Monthly plans are no longer sold on chutes.ai' (stored Plus $10 and Pro $20 rows kept from 2026-10-03, marked unverified, never ranked); no free tier. Matrix moves — re-curl before commits.","strengths":["Aggressive open-model inference pricing","Serverless chute deploys for custom models","Useful overflow capacity for batch agents"],"weaknesses":["Live token matrix still volatile vs Fireworks/Together","Support/SLA thinner for enterprise","Decentralized TEE marketplace operational risk"],"tags":["open-model","serverless","budget","inference","gpu"],"scores":{"capabilityFit":7.4,"pricePerformance":9.2,"latencyReliability":6.8,"developerDx":7.6,"privacyPosture":5.8,"lockInRisk":4,"maturity":6.2,"overall":7.29,"kind":"curated_estimate","confidence":"high","asOf":"2026-09-14","sources":["Chutes pricing: https://chutes.ai/pricing","Chutes starter guide: https://chutes.ai/docs/guides/starter-guide","Chutes: https://chutes.ai"],"notes":"Budget open-model overflow with a now-pinned public TEE matrix. Depot open-model production pick stays Fireworks; ultra-budget volume DeepInfra/Novita; use Chutes as capacity valve and re-check live rates."},"lastReviewed":"2026-09-14","status":"active"},{"id":"customer-io","slug":"customer-io","name":"Customer.io","website":"https://customer.io","tagline":"Best when agents need behavioral messaging + txn in one vendor. Prefer Resend for pure transactional DX…","description":"Best when agents need behavioral messaging + txn in one vendor. Prefer Resend for pure transactional DX; Loops for simple SaaS lifecycle. Strengths: Lifecycle + transactional in one product with strong event/segment model; Builder PAYG path aimed at AI coding agents ($0.40/1k); EU workspace option.","categoryIds":["cat-comms"],"jtbdIds":["marketing-email","transactional-email","notification-orchestration"],"regions":["global","us","eu"],"pricingNotes":"As of 2026-09-14: Essentials from ~$100/mo, Premium ~$1000/mo, Enterprise custom (customer.io/pricing OfferCatalog). Builder PAYG plan (docs): no monthly fee; free test to 10 verified recipients; go-live funds from $10 at $0.40 per 1,000 messages across email/push/in-app/webhooks (SMS/WhatsApp via Twilio). Transactional App API included — sends count toward allowances.","strengths":["Lifecycle + transactional in one product with strong event/segment model","Builder PAYG path aimed at AI coding agents ($0.40/1k)","EU workspace option"],"weaknesses":["Essentials/Premium seats heavier than Resend for pure txn email","SMS/WhatsApp still Twilio-piped on Builder","More CDP complexity than Loops/Resend"],"tags":["email","lifecycle","transactional","cdp","builder"],"scores":{"capabilityFit":8.6,"pricePerformance":7.8,"latencyReliability":8.6,"developerDx":8.4,"privacyPosture":8,"lockInRisk":4.8,"maturity":8.8,"overall":8.15,"kind":"curated_estimate","confidence":"high","asOf":"2026-09-14","sources":["Customer.io pricing: https://customer.io/pricing","Builder plan docs: https://docs.customer.io/accounts/billing/builder-plan/","Transactional email API: https://docs.customer.io/messaging/send/transactional/email/"],"notes":"Best when agents need behavioral messaging + txn in one vendor. Prefer Resend for pure transactional DX; Loops for simple SaaS lifecycle."},"lastReviewed":"2026-09-14","status":"active"},{"id":"amazon-ses","slug":"amazon-ses","name":"Amazon SES","website":"https://aws.amazon.com/ses/","tagline":"Budget transactional pipe when you already live on AWS and own deliverability. Default agents should still…","description":"Budget transactional pipe when you already live on AWS and own deliverability. Default agents should still start on Resend. Strengths: Lowest list $/1k among major ESP APIs; Deep AWS IAM/region/compliance story; Scales to huge volumes.","categoryIds":["cat-comms"],"jtbdIds":["transactional-email"],"regions":["global","us","eu","ap"],"pricingNotes":"As of 2026-09-14 (aws.amazon.com/ses/pricing): À la carte outbound $0.10 / 1,000 emails + $0.12/GB attachment data; inbound $0.10/1k. Plan tiers Essentials/Pro/Enterprise (e.g. Essentials ~$0.16/1k in 0–10M band) — can switch to à la carte in console. Cheapest hyperscaler hammer; deliverability/ops on you.","strengths":["Lowest list $/1k among major ESP APIs","Deep AWS IAM/region/compliance story","Scales to huge volumes"],"weaknesses":["DIY deliverability, templates, and observability vs Resend/Postmark","Plan vs à-la-carte confusion after 2026 packaging changes","Not agent-DX-first"],"tags":["email","aws","transactional","cheap"],"scores":{"capabilityFit":7.8,"pricePerformance":9.4,"latencyReliability":9,"developerDx":6.8,"privacyPosture":8.4,"lockInRisk":4.5,"maturity":9.5,"overall":8.19,"kind":"curated_estimate","confidence":"high","asOf":"2026-09-14","sources":["Amazon SES pricing: https://aws.amazon.com/ses/pricing/","Amazon SES: https://aws.amazon.com/ses/"],"notes":"Budget transactional pipe when you already live on AWS and own deliverability. Default agents should still start on Resend."},"lastReviewed":"2026-09-14","status":"active"},{"id":"stytch","slug":"stytch","name":"Stytch","website":"https://stytch.com","tagline":"Auth pick when agents/M2M + org SSO matter more than Clerk's consumer UI kit. Watch Twilio integration…","description":"Auth pick when agents/M2M + org SSO matter more than Clerk's consumer UI kit. Watch Twilio integration roadmap. Strengths: Native B2B multi-tenant + M2M tokens for agent workloads; Transparent $0 PAYG with SSO/SCIM included (not sales-gated basics); Strong fraud fingerprint add-on.","categoryIds":["cat-auth-payments"],"jtbdIds":["auth","enterprise-sso","user-management","agent-m2m-identity"],"regions":["global","us"],"pricingNotes":"As of 2026-09-14 (stytch.com/pricing B2B): Pay as you go $0 base — 10,000 MAU/agents included, unlimited orgs, 5 SSO/SCIM connections ($125/extra), 1,000 M2M tokens, 10k fingerprints then $0.005 each; brand customization $99/mo. Enterprise quote (99.99% SLA, HIPAA/BAA). Note: Stytch joined Twilio (identity layer) — product/brand still live.","strengths":["Native B2B multi-tenant + M2M tokens for agent workloads","Transparent $0 PAYG with SSO/SCIM included (not sales-gated basics)","Strong fraud fingerprint add-on"],"weaknesses":["Twilio acquisition → long-term packaging risk","Less prebuilt UI gravity than Clerk for consumer apps","HIPAA only on Enterprise"],"tags":["auth","b2b","sso","m2m","agents","twilio"],"scores":{"capabilityFit":8.6,"pricePerformance":8.8,"latencyReliability":8.4,"developerDx":8.8,"privacyPosture":8.2,"lockInRisk":4.2,"maturity":8,"overall":8.34,"kind":"curated_estimate","confidence":"high","asOf":"2026-09-14","sources":["Stytch pricing: https://stytch.com/pricing","Stytch: https://stytch.com"],"notes":"Auth pick when agents/M2M + org SSO matter more than Clerk's consumer UI kit. Watch Twilio integration roadmap."},"lastReviewed":"2026-09-14","status":"active"},{"id":"polar","slug":"polar","name":"Polar","website":"https://polar.sh","tagline":"Depot pick for ai-merchant-of-record. Prefer Stripe when you already have tax/ops staff; OpenMeter/Kong when…","description":"Depot pick for ai-merchant-of-record. Prefer Stripe when you already have tax/ops staff; OpenMeter/Kong when metering is internal-only. Strengths: Public MoR fee ladder (no sales-gated volume theater); Built for AI usage/seat/credit billing in one afternoon; OSS + strong agent/docs surface.","categoryIds":["cat-auth-payments"],"jtbdIds":["payments","billing","usage-metering","ai-merchant-of-record"],"regions":["global","us","eu"],"pricingNotes":"As of 2026-09-14 (polar.sh/resources/pricing): Merchant of Record. Starter free at 5%+$0.50/txn; Pro $20/mo + 3.8%+$0.40; Growth $100 + 3.6%+$0.35; Scale $400 + 3.4%+$0.30. Orgs before 2026-05-27 keep Early Member 4%+$0.40 (+0.5% subscriptions). +1.5% international cards. Usage metering for tokens/API/GPU events; LLMStrategy for Vercel AI SDK.","strengths":["Public MoR fee ladder (no sales-gated volume theater)","Built for AI usage/seat/credit billing in one afternoon","OSS + strong agent/docs surface"],"weaknesses":["MoR fees higher than raw Stripe for US-only simple SaaS","Younger than Stripe Billing at enterprise edge cases","Payouts still ride Stripe rails/fees"],"tags":["mor","billing","usage","ai","payments","oss"],"scores":{"capabilityFit":8.4,"pricePerformance":8.2,"latencyReliability":8,"developerDx":9,"privacyPosture":7.8,"lockInRisk":4,"maturity":7.2,"overall":8.05,"kind":"curated_estimate","confidence":"high","asOf":"2026-09-14","sources":["Polar pricing: https://polar.sh/resources/pricing","Polar fees docs: https://polar.sh/docs/merchant-of-record/fees","Polar: https://polar.sh"],"notes":"Depot pick for ai-merchant-of-record. Prefer Stripe when you already have tax/ops staff; OpenMeter/Kong when metering is internal-only."},"lastReviewed":"2026-09-14","status":"active"},{"id":"upstash-redis","slug":"upstash-redis","name":"Upstash Redis","website":"https://upstash.com/redis","tagline":"Default serverless-edge-redis. Use classic Redis Cloud when you need always-on fat memory / Redis modules…","description":"Default serverless-edge-redis. Use classic Redis Cloud when you need always-on fat memory / Redis modules; QStash for queues; Vector for embeddings. Strengths: HTTP Redis from Workers/Vercel/Lambda without connection pools; Agent-friendly free start-redis + MCP/skill surface; Per-request economics for bursty agents.","categoryIds":["cat-vector-memory"],"jtbdIds":["agent-memory-kv","serverless-edge-redis","rag-memory"],"regions":["global","us","eu"],"pricingNotes":"As of 2026-09-14 (upstash.com/pricing/redis): Free 256MB / 500K commands. PAYG $0.2 per 100K commands + $0.25/GB storage; fixed plans from $10/mo. Prod Pack +$200/db. HTTP + TCP; agents can POST https://upstash.com/start-redis for temp DBs. Separate from Upstash QStash/Vector/Box catalog entries.","strengths":["HTTP Redis from Workers/Vercel/Lambda without connection pools","Agent-friendly free start-redis + MCP/skill surface","Per-request economics for bursty agents"],"weaknesses":["Not a full Redis Enterprise feature parity story","Multi-product Upstash bill can fragment vs one Redis Cloud","Hot large working sets still favor classic Redis memory pricing"],"tags":["redis","serverless","edge","http","kv","agents"],"scores":{"capabilityFit":8.4,"pricePerformance":9,"latencyReliability":8.6,"developerDx":9.2,"privacyPosture":7.8,"lockInRisk":3.5,"maturity":8.6,"overall":8.49,"kind":"curated_estimate","confidence":"high","asOf":"2026-09-14","sources":["Upstash Redis pricing: https://upstash.com/pricing/redis","Upstash Redis: https://upstash.com/redis"],"notes":"Default serverless-edge-redis. Use classic Redis Cloud when you need always-on fat memory / Redis modules; QStash for queues; Vector for embeddings."},"lastReviewed":"2026-09-14","status":"active"},{"id":"upstash-vector","slug":"upstash-vector","name":"Upstash Vector","website":"https://upstash.com/vector","tagline":"Prototype/edge RAG peer when you already on Upstash. Prefer Qdrant/Pinecone for serious filtered production…","description":"Prototype/edge RAG peer when you already on Upstash. Prefer Qdrant/Pinecone for serious filtered production; Turbopuffer for cold billion-scale. Strengths: True serverless vector with clear request+storage card; HTTP/SDK DX matches rest of Upstash agent stack; Scale-to-zero free tier for prototypes.","categoryIds":["cat-vector-memory"],"jtbdIds":["prototype-rag","managed-vector","cheap-vector-scale","rag-memory"],"regions":["us","eu"],"pricingNotes":"As of 2026-09-14 (upstash.com/pricing/vector markdown): Free $0 — 200M vectors×dims, 10K queries/day, 1GB. PAYG $0.40 per 100K requests + $0.25/GB storage (≤50GB, 99.9% SLA). Fixed $60/mo (1M queries/day). Pro contact (100B scale). Bandwidth free to 200GB then $0.03/GB.","strengths":["True serverless vector with clear request+storage card","HTTP/SDK DX matches rest of Upstash agent stack","Scale-to-zero free tier for prototypes"],"weaknesses":["Feature depth behind Pinecone/Qdrant for hybrid filters","50GB PAYG cap before Pro","Fewer regions than hyperscaler vector SKUs"],"tags":["vector","serverless","rag","http"],"scores":{"capabilityFit":7.8,"pricePerformance":9,"latencyReliability":8.4,"developerDx":9,"privacyPosture":7.6,"lockInRisk":3.8,"maturity":7.8,"overall":8.15,"kind":"curated_estimate","confidence":"high","asOf":"2026-09-14","sources":["Upstash Vector pricing: https://upstash.com/pricing/vector","Upstash Vector: https://upstash.com/vector"],"notes":"Prototype/edge RAG peer when you already on Upstash. Prefer Qdrant/Pinecone for serious filtered production; Turbopuffer for cold billion-scale."},"lastReviewed":"2026-09-14","status":"active"},{"id":"upstash-box","slug":"upstash-box","name":"Upstash Box","website":"https://upstash.com/docs/box/overall/quickstart","tagline":"Interesting E2B peer for freezeable agent computers. Keep medium until preview exits; production default…","description":"Interesting E2B peer for freezeable agent computers. Keep medium until preview exits; production default remains E2B. Strengths: Freeze/resume agent computers with filesystem+shell; Parallel multi-agent fan-out with clear CPU-hour pricing; Native coding-agent harnesses.","categoryIds":["cat-sandboxes-browsers","cat-cpu-compute"],"jtbdIds":["agent-sandbox","code-interpreter","persistent-dev-env"],"regions":["global","us"],"pricingNotes":"As of 2026-09-14 (upstash.com/pricing/box): Still developer preview. Free: 10 concurrent, 5 CPU-hrs/mo, $1 LLM. PAYG Small/Med/Large **$0.10/$0.20/$0.40** per active CPU-hour; storage $0.1/GB-mo; Keep Alive $8/$16/$32/mo; PAYG LLM budget $100/mo + BYOK. Soft quota ~1000 concurrent. Re-pin when GA.","strengths":["Freeze/resume agent computers with filesystem+shell","Parallel multi-agent fan-out with clear CPU-hour pricing","Native coding-agent harnesses"],"weaknesses":["Still developer preview — APIs/pricing may move","Not as mature as E2B for production sandbox fleets","LLM token caps on free/PAYG unless BYOK"],"tags":["sandbox","agent-computer","preview","cpu"],"scores":{"capabilityFit":8,"pricePerformance":8.6,"latencyReliability":7.2,"developerDx":8.8,"privacyPosture":7.4,"lockInRisk":4.2,"maturity":5.8,"overall":7.67,"kind":"curated_estimate","confidence":"medium","asOf":"2026-09-14","sources":["Upstash Box pricing: https://upstash.com/pricing/box","Upstash Box quickstart: https://upstash.com/docs/box/overall/quickstart"],"notes":"Interesting E2B peer for freezeable agent computers. Keep medium until preview exits; production default remains E2B."},"lastReviewed":"2026-09-14","status":"active"},{"id":"wasabi","slug":"wasabi","name":"Wasabi Hot Cloud Storage","website":"https://wasabi.com","tagline":"Cheap hot archive peer when R2/Tigris edge story is not required. Depot cheap-egress still defaults to R2.","description":"Cheap hot archive peer when R2/Tigris edge story is not required. Depot cheap-egress still defaults to R2. Strengths: Predictable $/TB with zero egress/API request fees; S3-compatible for agent artifact dumps; Multi-region footprint.","categoryIds":["cat-object-storage"],"jtbdIds":["object-storage","cheap-egress-storage"],"regions":["global","us","eu","ap"],"pricingNotes":"As of 2026-09-14 (wasabi.com/pricing): Hot Cloud Storage from $7.99/TB-month PAYG — no egress fees, no API request fees (watch reserved-capacity and 2026 price-revision notes). Reserved capacity discounts. S3-compatible.","strengths":["Predictable $/TB with zero egress/API request fees","S3-compatible for agent artifact dumps","Multi-region footprint"],"weaknesses":["Less edge co-location magic than R2/Tigris","Minimum retention / control-plane quirks historically — read ToS","Not a CDN"],"tags":["s3","object-storage","egress-free","hot"],"scores":{"capabilityFit":7.8,"pricePerformance":9,"latencyReliability":8.4,"developerDx":7.8,"privacyPosture":7.6,"lockInRisk":3.2,"maturity":8.8,"overall":8.12,"kind":"curated_estimate","confidence":"high","asOf":"2026-09-14","sources":["Wasabi pricing: https://wasabi.com/pricing","Wasabi: https://wasabi.com"],"notes":"Cheap hot archive peer when R2/Tigris edge story is not required. Depot cheap-egress still defaults to R2."},"lastReviewed":"2026-09-14","status":"active"},{"id":"serpapi","slug":"serpapi","name":"SerpApi","website":"https://serpapi.com","tagline":"Classic SERP API when you need engine breadth. Prefer Serper for budget Google-only; Exa/Tavily/You for…","description":"Classic SERP API when you need engine breadth. Prefer Serper for budget Google-only; Exa/Tavily/You for agent-native search. Strengths: Broadest mature SERP engine coverage; Stable JSON schemas agents already know; Clear monthly search tiers.","categoryIds":["cat-search-scrape"],"jtbdIds":["serp-scrape","agent-search"],"regions":["global","us"],"pricingNotes":"As of 2026-09-14 (serpapi.com/pricing): Free $0 — 250 searches/mo. Starter $25/1k, Developer $75/5k, Production $150/15k, Big Data $275/30k (month-to-month). Structured Google/Bing/Amazon/etc SERP JSON. Legal Shield add-on available.","strengths":["Broadest mature SERP engine coverage","Stable JSON schemas agents already know","Clear monthly search tiers"],"weaknesses":["Not LLM-answer-shaped like Exa/Tavily/You.com","$/search higher than Serper at similar volumes","ToS/compliance diligence still required for scraping"],"tags":["serp","google","search","json"],"scores":{"capabilityFit":8.6,"pricePerformance":7.4,"latencyReliability":8.8,"developerDx":8.6,"privacyPosture":7,"lockInRisk":3.5,"maturity":9,"overall":8.15,"kind":"curated_estimate","confidence":"high","asOf":"2026-09-14","sources":["SerpApi pricing: https://serpapi.com/pricing","SerpApi: https://serpapi.com"],"notes":"Classic SERP API when you need engine breadth. Prefer Serper for budget Google-only; Exa/Tavily/You for agent-native search."},"lastReviewed":"2026-09-14","status":"active"},{"id":"ideogram","slug":"ideogram","name":"Ideogram","website":"https://ideogram.ai","tagline":"Use when agents must render readable text in images. Fal remains default multi-model media router.","description":"Use when agents must render readable text in images. Fal remains default multi-model media router. Strengths: Excellent typography / text rendering in images; Simple per-image API tiers; No forced subscription for API.","categoryIds":["cat-media-gen"],"jtbdIds":["image-generation","media-generation","text-in-image-generation"],"regions":["global","us"],"pricingNotes":"As of 2026-09-14: Consumer Plus ~$20/mo (site). API pay-as-you-go per image (ideogram.ai/api-pricing): 4.0 Turbo ~$0.03, Default ~$0.06, Quality ~$0.10; volume via sales. Strong text-in-image reputation.","strengths":["Excellent typography / text rendering in images","Simple per-image API tiers","No forced subscription for API"],"weaknesses":["Narrower model zoo than Fal/Replicate","API pricing page sometimes bot-friction","Less video/3D surface"],"tags":["image","typography","text-in-image","api"],"scores":{"capabilityFit":8.4,"pricePerformance":8,"latencyReliability":8,"developerDx":8.2,"privacyPosture":7,"lockInRisk":3.8,"maturity":7.8,"overall":7.88,"kind":"curated_estimate","confidence":"high","asOf":"2026-09-14","sources":["Ideogram API pricing: https://ideogram.ai/api-pricing","Ideogram pricing: https://ideogram.ai/pricing","Ideogram: https://ideogram.ai"],"notes":"Use when agents must render readable text in images. Fal remains default multi-model media router."},"lastReviewed":"2026-09-14","status":"active"},{"id":"recraft","slug":"recraft","name":"Recraft","website":"https://www.recraft.ai","tagline":"Best when agents need SVG/brand-consistent assets. Fal for general media zoo; BFL for FLUX-first.","description":"Best when agents need SVG/brand-consistent assets. Fal for general media zoo; BFL for FLUX-first. Strengths: First-class vector + brand-style image API; Transparent prepaid unit matrix; Strong design/branding agent fit.","categoryIds":["cat-media-gen"],"jtbdIds":["image-generation","media-generation"],"regions":["global","us","eu"],"pricingNotes":"As of 2026-09-14 (recraft.ai docs API pricing): Prepaid API units $1 = 1,000 units (never expire). V4.1 raster $0.035 (35u) / Pro $0.21 (210u); V4.1 vector $0.08 / Pro Vector $0.30; V3 ops ~$0.04 raster / $0.08 vector; bg removal $0.01; crisp upscale $0.004; creative upscale $0.25.","strengths":["First-class vector + brand-style image API","Transparent prepaid unit matrix","Strong design/branding agent fit"],"weaknesses":["Prepaid packs (no pure PAYG surprise billing)","Less 'fire every HF model' breadth than Fal","Consumer Free plan ownership differs from paid"],"tags":["image","vector","brand","design","api"],"scores":{"capabilityFit":8.6,"pricePerformance":8.2,"latencyReliability":8.2,"developerDx":8.6,"privacyPosture":7.4,"lockInRisk":4,"maturity":7.6,"overall":8.07,"kind":"curated_estimate","confidence":"high","asOf":"2026-09-14","sources":["Recraft API pricing: https://www.recraft.ai/docs/api-reference/pricing","Recraft: https://www.recraft.ai"],"notes":"Best when agents need SVG/brand-consistent assets. Fal for general media zoo; BFL for FLUX-first."},"lastReviewed":"2026-09-14","status":"active"},{"id":"bird","slug":"bird","name":"Bird (ex-MessageBird)","website":"https://bird.com","tagline":"WhatsApp/SMS peer especially for EU (ex-MessageBird). Twilio remains default SMS/voice toolkit; Bird when…","description":"WhatsApp/SMS peer especially for EU (ex-MessageBird). Twilio remains default SMS/voice toolkit; Bird when public WhatsApp category rates + no-seat message card matter. Strengths: Public WhatsApp category rates with Meta fee included; EU-rooted CPaaS alternative to Twilio; Omnichannel customer engagement suite.","categoryIds":["cat-comms"],"jtbdIds":["whatsapp","sms","notification-orchestration"],"regions":["global","eu","us"],"pricingNotes":"As of 2026-09-14 (bird.com WhatsApp pricing): Usage-based WhatsApp with Meta fee included and explicitly **no seat fee** / no annual commit on the message card — US marketing $0.030, utility/auth $0.0084, service $0.0050 per delivered msg. Docs still describe Bird Processing + passthrough for some legacy/subscription paths — confirm wallet vs seat packaging in console for omnichannel suites.","strengths":["Public WhatsApp category rates with Meta fee included","EU-rooted CPaaS alternative to Twilio","Omnichannel customer engagement suite"],"weaknesses":["Legacy docs still mention processing/passthrough vs all-in message card — reconcile account type","Rebrand MessageBird→Bird confuses older integrations","Agent toolkit thinner than Twilio+Courier stacks"],"tags":["whatsapp","sms","cpaas","messagebird"],"scores":{"capabilityFit":8.2,"pricePerformance":7.8,"latencyReliability":8.4,"developerDx":7.6,"privacyPosture":8,"lockInRisk":4.8,"maturity":8.6,"overall":7.88,"kind":"curated_estimate","confidence":"high","asOf":"2026-09-14","sources":["Bird WhatsApp pricing (US): https://bird.com/en-us/pricing/whatsapp","Bird WhatsApp pricing: https://bird.com/en-au/pricing/whatsapp","Bird pricing docs: https://docs.bird.com/applications/bird-documentation/bird-pricing"],"notes":"WhatsApp/SMS peer especially for EU (ex-MessageBird). Twilio remains default SMS/voice toolkit; Bird when public WhatsApp category rates + no-seat message card matter."},"lastReviewed":"2026-09-14","status":"active"},{"id":"friendli","slug":"friendli","name":"FriendliAI","website":"https://friendli.ai","tagline":"Open-model production peer when latency/tool-calling matter. Fireworks remains depot open-model default…","description":"Open-model production peer when latency/tool-calling matter. Fireworks remains depot open-model default; dedicated GPU step-up live 2026-10-01 — size reserved endpoints on the new $/hr card. Strengths: Production open-weight inference with strong latency/tool-calling claims; Clear dedicated GPU $/hr card (post-2026-10-01 step-up); OpenRouter benchmark presence.","categoryIds":["cat-inference-apis","cat-gpu-cloud"],"jtbdIds":["open-model-inference","ultra-low-latency","production-inference-fleet","serverless-gpu"],"regions":["global","us","ap"],"pricingNotes":"As of 2026-10-01 (friendli.ai/pricing live): Dedicated Endpoints on-demand step-up **is live** — A100 **$4.0**/hr, H100 **$5.0**/hr, H200 **$7.0**/hr, B200 **$9.0**/hr, B300 $12.0/hr (was through-Sep-30: A100 $2.9, H100 $3.9, H200 $4.5, B200 $8.9, B300 $12.0). Billed per second; no idle/startup charge. Model API token examples on page (GLM/DeepSeek/MiniMax/whisper). SOC2 + HIPAA.","strengths":["Production open-weight inference with strong latency/tool-calling claims","Clear dedicated GPU $/hr card (post-2026-10-01 step-up)","OpenRouter benchmark presence"],"weaknesses":["Model API token matrix less scrape-friendly than Fireworks/Together","Dedicated rates stepped up 2026-10-01 (A100/H100/H200 meaningful; B200 slight)","Smaller Western mindshare vs Fireworks/Groq"],"tags":["inference","open-weights","dedicated-gpu","low-latency"],"scores":{"capabilityFit":8.4,"pricePerformance":8,"latencyReliability":8.4,"developerDx":8.2,"privacyPosture":8.2,"lockInRisk":4,"maturity":7.6,"overall":8.03,"kind":"curated_estimate","confidence":"high","asOf":"2026-10-01","sources":["FriendliAI pricing: https://friendli.ai/pricing","FriendliAI: https://friendli.ai"],"notes":"Open-model production peer when latency/tool-calling matter. Fireworks remains depot open-model default; dedicated GPU step-up live 2026-10-01 — size reserved endpoints on the new $/hr card."},"lastReviewed":"2026-10-01","status":"active"},{"id":"typesense-cloud","slug":"typesense-cloud","name":"Typesense Cloud","website":"https://cloud.typesense.org","tagline":"Hybrid search peer when you want Algolia-like DX without per-request fees. Sample $0.03/hr floor pinned…","description":"Hybrid search peer when you want Algolia-like DX without per-request fees. Sample $0.03/hr floor pinned; always export the HA SKU you will actually run. Strengths: Predictable cluster economics (no per-query surprise); Excellent typo-tolerant keyword + vector hybrid; OSS escape hatch.","categoryIds":["cat-search-scrape","cat-vector-memory"],"jtbdIds":["hybrid-search-rag","neural-search","enterprise-search"],"regions":["global","us","eu"],"pricingNotes":"As of 2026-09-14: Hourly hardware clusters via cloud.typesense.org calculator (no per-search fees). Sample entry SKU ~0.5 GB RAM / 2 burst vCPU ≈ **$0.03/hr** (~$21.60/720h-mo); HA multi-node and higher RAM multiply. Bandwidth ~$0.09–$0.12/GB class on secondary trackers; 10 GB free on eligible accounts. OSS self-host free. Export exact HA quote from configurator before commits.","strengths":["Predictable cluster economics (no per-query surprise)","Excellent typo-tolerant keyword + vector hybrid","OSS escape hatch"],"weaknesses":["Still configurator-based (not one flat SaaS card)","You operate capacity vs fully serverless Algolia/Upstash","Smaller ecosystem than Elasticsearch/Algolia"],"tags":["search","hybrid","oss","vector"],"scores":{"capabilityFit":8.4,"pricePerformance":8.6,"latencyReliability":8.4,"developerDx":8.6,"privacyPosture":7.8,"lockInRisk":3,"maturity":8.2,"overall":8.29,"kind":"curated_estimate","confidence":"high","asOf":"2026-09-14","sources":["Typesense Cloud pricing calculator: https://cloud.typesense.org/pricing/calculator","Typesense Cloud pricing: https://cloud.typesense.org/pricing","Typesense Cloud billing help: https://cloud-help.typesense.org/article/billing-process"],"notes":"Hybrid search peer when you want Algolia-like DX without per-request fees. Sample $0.03/hr floor pinned; always export the HA SKU you will actually run."},"lastReviewed":"2026-09-14","status":"active"},{"id":"algolia","slug":"algolia","name":"Algolia","website":"https://www.algolia.com","tagline":"Enterprise product-search default. Prefer Typesense/OpenSearch when $/QPS dominates; vector DBs when pure…","description":"Enterprise product-search default. Prefer Typesense/OpenSearch when $/QPS dominates; vector DBs when pure embedding RAG. Strengths: Gold-standard hosted search DX and relevance tooling; Global edge search latency; Mature ecommerce/agent retrieval patterns.","categoryIds":["cat-search-scrape"],"jtbdIds":["enterprise-search","neural-search","hybrid-search-rag"],"regions":["global","us","eu"],"pricingNotes":"As of 2026-09-14 (algolia.com/pricing): Pay-as-you-go — start free with 10K search requests/mo then ~$0.50–$1.75 per 1k depending on feature pack (Grow keyword vs AI Search tiers). Higher Medium/High usage and Enterprise support plans. Exact AI feature mix moves — pin calculator.","strengths":["Gold-standard hosted search DX and relevance tooling","Global edge search latency","Mature ecommerce/agent retrieval patterns"],"weaknesses":["Can get expensive vs Typesense/self-host at high QPS","AI Search SKUs add pricing complexity","Overkill for simple RAG vector-only"],"tags":["search","enterprise","neural","hosted"],"scores":{"capabilityFit":9,"pricePerformance":6.8,"latencyReliability":9.2,"developerDx":9,"privacyPosture":8,"lockInRisk":5.5,"maturity":9.5,"overall":8.27,"kind":"curated_estimate","confidence":"high","asOf":"2026-09-14","sources":["Algolia pricing: https://www.algolia.com/pricing","Algolia: https://www.algolia.com"],"notes":"Enterprise product-search default. Prefer Typesense/OpenSearch when $/QPS dominates; vector DBs when pure embedding RAG."},"lastReviewed":"2026-09-14","status":"active"},{"id":"diffbot","slug":"diffbot","name":"Diffbot","website":"https://www.diffbot.com","tagline":"Structured extract / KG peer. Firecrawl remains llm-ready-extract default; Diffbot when you need schema'd…","description":"Structured extract / KG peer. Firecrawl remains llm-ready-extract default; Diffbot when you need schema'd entities/KG. Strengths: Structured page Extract + Knowledge Graph in one vendor; Clear free/Startup credit economics; Less brittle than DIY CSS scrapers for product pages.","categoryIds":["cat-search-scrape"],"jtbdIds":["llm-ready-extract","web-scrape","knowledge-graph-extract"],"regions":["global","us"],"pricingNotes":"As of 2026-09-14 (diffbot.com/pricing): Free $0 — 9,500 credits/mo, 5 rpm. Startup $299/mo for production. Credits: Extract 1 page = 1 credit (2 via DC proxy); NLP doc = 1; Knowledge Graph queries metered. Overage pro-rata per-credit. Higher tiers via sales.","strengths":["Structured page Extract + Knowledge Graph in one vendor","Clear free/Startup credit economics","Less brittle than DIY CSS scrapers for product pages"],"weaknesses":["Not as LLM-markdown-native as Firecrawl/Jina Reader","$299 Startup jump from Free","KG coverage/licensing diligence needed"],"tags":["extract","knowledge-graph","crawl","nlp"],"scores":{"capabilityFit":8.6,"pricePerformance":7.6,"latencyReliability":8.4,"developerDx":8,"privacyPosture":7.2,"lockInRisk":4.5,"maturity":8.8,"overall":7.96,"kind":"curated_estimate","confidence":"high","asOf":"2026-09-14","sources":["Diffbot pricing: https://www.diffbot.com/pricing/","Diffbot: https://www.diffbot.com"],"notes":"Structured extract / KG peer. Firecrawl remains llm-ready-extract default; Diffbot when you need schema'd entities/KG."},"lastReviewed":"2026-09-14","status":"active"},{"id":"you-com","slug":"you-com","name":"You.com API","website":"https://you.com","tagline":"Agent search peer with Answer/Research SKUs. Exa remains depot agent-search pick; You.com strong when ZDR +…","description":"Agent search peer with Answer/Research SKUs. Exa remains depot agent-search pick; You.com strong when ZDR + machine payments matter. Strengths: Agent-native search/answer/research APIs with clear $/1k; Machine payments path for autonomous agents; ZDR + SOC2 story.","categoryIds":["cat-search-scrape","cat-inference-apis"],"jtbdIds":["agent-search","search-augmented-llm","web-grounded-qa"],"regions":["global","us"],"pricingNotes":"As of 2026-09-14 (you.com/pricing + docs): $100 free credit; free 100 Web Search queries/day without key. Web Search API $5.00/1k calls; Contents API $1.00/1k pages; Answer API $5.00/1k; Research API tiers from Lite→Exhaustive (docs show ~$12–$1,200/1k depending on depth). SOC2; ZDR available; x402/MPP machine payments supported.","strengths":["Agent-native search/answer/research APIs with clear $/1k","Machine payments path for autonomous agents","ZDR + SOC2 story"],"weaknesses":["Research Exhaustive tiers get expensive fast","Younger than Exa/Tavily in agent mindshare","Contents vs Search packaging easy to mis-budget"],"tags":["search","agents","research","zdr","answer"],"scores":{"capabilityFit":8.4,"pricePerformance":7.8,"latencyReliability":8.2,"developerDx":8.8,"privacyPosture":8.4,"lockInRisk":3.8,"maturity":7.4,"overall":8.07,"kind":"curated_estimate","confidence":"high","asOf":"2026-09-14","sources":["You.com pricing: https://you.com/pricing","You.com billing docs: https://you.com/docs/administration/billing","You.com: https://you.com"],"notes":"Agent search peer with Answer/Research SKUs. Exa remains depot agent-search pick; You.com strong when ZDR + machine payments matter."},"lastReviewed":"2026-09-14","status":"active"},{"id":"google-cloud-storage","slug":"google-cloud-storage","name":"Google Cloud Storage","website":"https://cloud.google.com/storage","tagline":"Use inside GCP estates. Depot object-storage/cheap-egress defaults remain R2.","description":"Use inside GCP estates. Depot object-storage/cheap-egress defaults remain R2. Strengths: Global durability and IAM integration with GCP agent stacks; Tight Vertex/BigQuery adjacency; Mature lifecycle/autoclass.","categoryIds":["cat-object-storage"],"jtbdIds":["object-storage"],"regions":["global","us","eu","ap"],"pricingNotes":"As of 2026-09-14: Standard/Nearline/Coldline/Archive storage + ops + egress (egress often dominates agent artifact serving). Autoclass available. Prefer when already on GCP; for zero-egress public serving prefer R2/Tigris/Wasabi/B2.","strengths":["Global durability and IAM integration with GCP agent stacks","Tight Vertex/BigQuery adjacency","Mature lifecycle/autoclass"],"weaknesses":["Egress fees punish public agent artifact delivery","Pricing complexity vs R2 flat story","Lock-in with GCP IAM patterns"],"tags":["gcs","object-storage","gcp"],"scores":{"capabilityFit":8.8,"pricePerformance":7,"latencyReliability":9.4,"developerDx":7.8,"privacyPosture":8.6,"lockInRisk":5.5,"maturity":9.6,"overall":8.18,"kind":"curated_estimate","confidence":"high","asOf":"2026-09-14","sources":["Cloud Storage pricing: https://cloud.google.com/storage/pricing","Google Cloud Storage: https://cloud.google.com/storage"],"notes":"Use inside GCP estates. Depot object-storage/cheap-egress defaults remain R2."},"lastReviewed":"2026-09-14","status":"active"},{"id":"azure-blob","slug":"azure-blob","name":"Azure Blob Storage","website":"https://azure.microsoft.com/products/storage/blobs","tagline":"Azure-estate default object store. Prefer R2/Tigris when egress-free public URLs matter.","description":"Azure-estate default object store. Prefer R2/Tigris when egress-free public URLs matter. Strengths: Enterprise Azure compliance + Private Link story; Native fit with Azure OpenAI agent stacks; Hierarchical namespace / Data Lake options.","categoryIds":["cat-object-storage"],"jtbdIds":["object-storage"],"regions":["global","us","eu","ap"],"pricingNotes":"As of 2026-09-14: Hot/Cool/Cold/Archive tiers + transactions + egress. Azure OpenAI / AI Foundry adjacency. Hot storage competitive; internet egress still the gotcha for agent CDN-less serving.","strengths":["Enterprise Azure compliance + Private Link story","Native fit with Azure OpenAI agent stacks","Hierarchical namespace / Data Lake options"],"weaknesses":["Egress and tiering complexity","Weaker indie DX than R2/S3","Lock-in with Azure IAM"],"tags":["azure","blob","object-storage"],"scores":{"capabilityFit":8.6,"pricePerformance":7,"latencyReliability":9.4,"developerDx":7.4,"privacyPosture":8.8,"lockInRisk":5.8,"maturity":9.6,"overall":8.06,"kind":"curated_estimate","confidence":"high","asOf":"2026-09-14","sources":["Azure Blob Storage pricing: https://azure.microsoft.com/pricing/details/storage/blobs/","Azure Blob Storage: https://azure.microsoft.com/products/storage/blobs"],"notes":"Azure-estate default object store. Prefer R2/Tigris when egress-free public URLs matter."},"lastReviewed":"2026-09-14","status":"active"},{"id":"planetscale","slug":"planetscale","name":"PlanetScale","website":"https://planetscale.com","tagline":"Serverless DB peer when branching DX matters. Prefer Neon/Supabase for default Postgres+pgvector agents…","description":"Serverless DB peer when branching DX matters. Prefer Neon/Supabase for default Postgres+pgvector agents; PlanetScale when Vitess/MySQL or schema-branch workflows dominate. Strengths: Best-in-class database branching / nonblocking schema deploy DX; Horizontal MySQL heritage + Postgres path; Good fit for agents that migrate schemas often.","categoryIds":["cat-cpu-compute","cat-vector-memory"],"jtbdIds":["serverless-db","app-backend","postgres-vector"],"regions":["global","us","eu"],"pricingNotes":"As of 2026-09-14 (planetscale.com/pricing.md, AWS us-east-1): Postgres single-node from **$5/mo** (PS-5); HA 3-node from **$15/mo** (PS-5); PS-160 ARM HA $286; Vitess HA PS-10 $39+. Metal NVMe SKUs from ~$50–$60 (M-10). Storage/backups/egress/PgBouncer extra. Scaler Pro renamed Base. Machine-readable SQL catalog + region query params — re-export before commits.","strengths":["Best-in-class database branching / nonblocking schema deploy DX","Horizontal MySQL heritage + Postgres path","Good fit for agents that migrate schemas often"],"weaknesses":["SKU matrix large (Postgres/Vitess/Neki/Metal) — always query pricing.md/SQL catalog","Not a dedicated vector DB","Can be pricier than Neon/Supabase for simple Postgres"],"tags":["mysql","postgres","serverless","branching"],"scores":{"capabilityFit":8.6,"pricePerformance":7.4,"latencyReliability":8.8,"developerDx":9.2,"privacyPosture":8,"lockInRisk":4.8,"maturity":8.8,"overall":8.23,"kind":"curated_estimate","confidence":"high","asOf":"2026-09-14","sources":["PlanetScale pricing.md: https://planetscale.com/pricing.md","PlanetScale pricing: https://planetscale.com/pricing","PlanetScale Postgres pricing docs: https://planetscale.com/docs/postgres/pricing"],"notes":"Serverless DB peer when branching DX matters. Prefer Neon/Supabase for default Postgres+pgvector agents; PlanetScale when Vitess/MySQL or schema-branch workflows dominate."},"lastReviewed":"2026-09-14","status":"active"},{"id":"telnyx","slug":"telnyx","name":"Telnyx","website":"https://telnyx.com","tagline":"SMS/voice peer when $/part and published carrier fees matter. Twilio remains default toolkit gravity; Bird…","description":"SMS/voice peer when $/part and published carrier fees matter. Twilio remains default toolkit gravity; Bird for WhatsApp-first EU; Plivo for voice-AI packaging. Strengths: Lowest public US SMS list vs Twilio/Vonage peers; Carrier fees published separately (no opaque markup theater); Same account for SMS/voice/SIP/WhatsApp useful for voice agents.","categoryIds":["cat-comms"],"jtbdIds":["sms","whatsapp","voice","voice-agents"],"regions":["global","us","eu"],"pricingNotes":"As of 2026-09-14 (telnyx.com/pricing/messaging): US local SMS from **$0.004/part** + published carrier passthrough; $0 platform fee; volume tiers after 100M parts. MMS from $0.015/part. Machine-readable GET /v2/public/pricing. WhatsApp BSP + Email API beta on same account.","strengths":["Lowest public US SMS list vs Twilio/Vonage peers","Carrier fees published separately (no opaque markup theater)","Same account for SMS/voice/SIP/WhatsApp useful for voice agents"],"weaknesses":["Carrier passthrough still dominates all-in cost","10DLC/compliance ops still required","Notification orchestration thinner than Courier/Knock"],"tags":["sms","cpaas","voice","whatsapp","sip"],"scores":{"capabilityFit":8.4,"pricePerformance":9,"latencyReliability":8.6,"developerDx":8.8,"privacyPosture":8.2,"lockInRisk":3.8,"maturity":8.4,"overall":8.42,"kind":"curated_estimate","confidence":"high","asOf":"2026-09-14","sources":["Telnyx messaging pricing: https://telnyx.com/pricing/messaging","Telnyx: https://telnyx.com"],"notes":"SMS/voice peer when $/part and published carrier fees matter. Twilio remains default toolkit gravity; Bird for WhatsApp-first EU; Plivo for voice-AI packaging."},"lastReviewed":"2026-09-14","status":"active"},{"id":"plivo","slug":"plivo","name":"Plivo","website":"https://www.plivo.com","tagline":"CPaaS peer especially when Voice AI Agents + SIP streaming matter. Depot voice-agent platform pick stays…","description":"CPaaS peer especially when Voice AI Agents + SIP streaming matter. Depot voice-agent platform pick stays Vapi; SMS default Twilio with Telnyx for $/part. Strengths: Voice AI agent SKU with LiveKit/Pipecat SIP streaming; Clear SMS + WhatsApp + voice card on one page; Good India/US coverage heritage.","categoryIds":["cat-comms"],"jtbdIds":["sms","whatsapp","voice","voice-agent-platform"],"regions":["global","us","in"],"pricingNotes":"As of 2026-09-14 (plivo.com/pricing): PAYG with $10 free credits. SMS from **$0.0077**; WhatsApp from $0.0037/msg; voice calls from $0.0010/min outbound; Voice AI Agents **$0.03/min** excluding telephony. Enterprise from $1,000/mo for 190+ countries + compliance.","strengths":["Voice AI agent SKU with LiveKit/Pipecat SIP streaming","Clear SMS + WhatsApp + voice card on one page","Good India/US coverage heritage"],"weaknesses":["SMS list higher than Telnyx","Enterprise floor $1k/mo for full international","Smaller ecosystem than Twilio"],"tags":["sms","voice-ai","whatsapp","cpaas"],"scores":{"capabilityFit":8.2,"pricePerformance":8.4,"latencyReliability":8.4,"developerDx":8.4,"privacyPosture":7.8,"lockInRisk":4,"maturity":8.2,"overall":8.1,"kind":"curated_estimate","confidence":"high","asOf":"2026-09-14","sources":["Plivo pricing: https://www.plivo.com/pricing/","Plivo: https://www.plivo.com"],"notes":"CPaaS peer especially when Voice AI Agents + SIP streaming matter. Depot voice-agent platform pick stays Vapi; SMS default Twilio with Telnyx for $/part."},"lastReviewed":"2026-09-14","status":"active"},{"id":"mailgun","slug":"mailgun","name":"Mailgun","website":"https://www.mailgun.com","tagline":"Txn email peer when EU residency or deliverability ops matter. Depot default remains Resend; Postmark for…","description":"Txn email peer when EU residency or deliverability ops matter. Depot default remains Resend; Postmark for crisis deliverability; SES for AWS-cheap pipe. Strengths: Battle-tested transactional API + SMTP; Explicit US/EU region choice; Deliverability tooling (validation, dedicated IPs) at Scale.","categoryIds":["cat-comms"],"jtbdIds":["transactional-email","marketing-email"],"regions":["global","us","eu"],"pricingNotes":"As of 2026-09-14 (mailgun.com/pricing): Free 100 emails/day; Basic from **$15/mo** (10k); Foundation **$35/mo** (50k); Scale **$90/mo** (100k). Overage from ~$1.10–$1.80/1k. US or EU data residency. Dedicated IPs extra (~$59/IP/mo class).","strengths":["Battle-tested transactional API + SMTP","Explicit US/EU region choice","Deliverability tooling (validation, dedicated IPs) at Scale"],"weaknesses":["Price often higher than Resend for simple txn mail","Sinch ownership / console UX heavier than indie APIs","Marketing features overlap poorly with Loops/Customer.io"],"tags":["email","transactional","smtp","eu"],"scores":{"capabilityFit":8.6,"pricePerformance":7.2,"latencyReliability":9,"developerDx":8.2,"privacyPosture":8.4,"lockInRisk":4.2,"maturity":9.2,"overall":8.19,"kind":"curated_estimate","confidence":"high","asOf":"2026-09-14","sources":["Mailgun pricing: https://www.mailgun.com/pricing/","Mailgun: https://www.mailgun.com"],"notes":"Txn email peer when EU residency or deliverability ops matter. Depot default remains Resend; Postmark for crisis deliverability; SES for AWS-cheap pipe."},"lastReviewed":"2026-09-14","status":"active"},{"id":"lemon-squeezy","slug":"lemon-squeezy","name":"Lemon Squeezy","website":"https://www.lemonsqueezy.com","tagline":"MoR peer to Polar/Paddle for digital products. Depot ai-merchant-of-record pick stays Polar for AI…","description":"MoR peer to Polar/Paddle for digital products. Depot ai-merchant-of-record pick stays Polar for AI usage/credit ladders; Lemon for classic digital-goods DX. Strengths: Simple MoR fee for digital/SaaS makers; License keys + checkout + tax handled; Fast indie onboarding vs building Stripe Tax.","categoryIds":["cat-auth-payments"],"jtbdIds":["payments","billing","ai-merchant-of-record"],"regions":["global","us"],"pricingNotes":"As of 2026-09-14 (lemonsqueezy.com/pricing): Merchant of Record at **5% + $0.50**/txn; $0/mo platform fee for ecommerce features. Edge-case adders (intl/PayPal/subscriptions) documented. Email marketing separate by subscriber count (free ≤500).","strengths":["Simple MoR fee for digital/SaaS makers","License keys + checkout + tax handled","Fast indie onboarding vs building Stripe Tax"],"weaknesses":["MoR fee higher than raw Stripe for US-only","Less AI-usage-metering native than Polar","Additional fee edge cases need reading the fee docs"],"tags":["mor","payments","billing","saas"],"scores":{"capabilityFit":8,"pricePerformance":7.8,"latencyReliability":8.4,"developerDx":8.6,"privacyPosture":7.6,"lockInRisk":4.5,"maturity":8,"overall":7.9,"kind":"curated_estimate","confidence":"high","asOf":"2026-09-14","sources":["Lemon Squeezy pricing: https://www.lemonsqueezy.com/pricing","Lemon Squeezy fees docs: https://docs.lemonsqueezy.com/help/getting-started/fees"],"notes":"MoR peer to Polar/Paddle for digital products. Depot ai-merchant-of-record pick stays Polar for AI usage/credit ladders; Lemon for classic digital-goods DX."},"lastReviewed":"2026-09-14","status":"active"},{"id":"paddle","slug":"paddle","name":"Paddle","website":"https://www.paddle.com","tagline":"Enterprise-leaning MoR peer. Prefer Polar for AI usage products; Paddle when you want mature SaaS MoR + buyer…","description":"Enterprise-leaning MoR peer. Prefer Polar for AI usage products; Paddle when you want mature SaaS MoR + buyer support. Strengths: Mature MoR for serious SaaS cross-border; Buyer billing support handled by Paddle; Strong subscription + tax story.","categoryIds":["cat-auth-payments"],"jtbdIds":["payments","billing","ai-merchant-of-record"],"regions":["global","us","eu"],"pricingNotes":"As of 2026-09-14 (paddle.com/pricing): All-in MoR **5% + $0.50** per Checkout txn; no monthly fee. Custom pricing for scale / sub-$10 products / invoicing. Includes tax remittance, fraud, billing support.","strengths":["Mature MoR for serious SaaS cross-border","Buyer billing support handled by Paddle","Strong subscription + tax story"],"weaknesses":["Heavier onboarding than Polar/Lemon for tiny agents","Less AI-token metering native than Polar/OpenMeter","MoR lock-in on checkout/tax flows"],"tags":["mor","payments","billing","saas","tax"],"scores":{"capabilityFit":8.6,"pricePerformance":7.6,"latencyReliability":8.8,"developerDx":8,"privacyPosture":8,"lockInRisk":5.5,"maturity":9,"overall":8.05,"kind":"curated_estimate","confidence":"high","asOf":"2026-09-14","sources":["Paddle pricing: https://www.paddle.com/pricing","Paddle: https://www.paddle.com"],"notes":"Enterprise-leaning MoR peer. Prefer Polar for AI usage products; Paddle when you want mature SaaS MoR + buyer support."},"lastReviewed":"2026-09-14","status":"active"},{"id":"svix","slug":"svix","name":"Svix","website":"https://www.svix.com","tagline":"Depot pick for outbound-webhooks. Use Hookdeck when the job is inbound event gateway/debugging; Courier/Knock…","description":"Depot pick for outbound-webhooks. Use Hookdeck when the job is inbound event gateway/debugging; Courier/Knock for multi-channel notify. Strengths: Purpose-built outbound webhook delivery (retries, portal, signatures); Generous free tier for agent products exposing webhooks; Ingest + Stream + Play in one platform.","categoryIds":["cat-comms","cat-cpu-compute"],"jtbdIds":["outbound-webhooks","notification-orchestration","event-driven-agent-jobs"],"regions":["global","us","eu"],"pricingNotes":"As of 2026-10-24 (svix.com/pricing, fetched 2026-10-24): Free 'from $ 0 per month' ('50 messages / second', '7-day payload retention', 'Up to 50k messages per month'). Basic 'from $ 20 per month' ('Uptime SLA of 99.9%', '200 messages / second', '30-day payload retention', '50k free messages a month'). Professional 'from $ 490 per month' ('Uptime SLA of 99.99%', '800 messages / second', '90-day payload retention', 'Unbranded application portal', 'Static source IPs'). Overage '$0.0001 per msg' on Basic and Professional; 'Messages filtered by Svix (e.g. when there are no endpoints) and retries are both free.' Enterprise custom ('Uptime SLA of 99.999%', 'On-prem options'). See plan limits (/api/estimate) for the encoded plans.","strengths":["Purpose-built outbound webhook delivery (retries, portal, signatures)","Generous free tier for agent products exposing webhooks","Ingest + Stream + Play in one platform"],"weaknesses":["Pro jumps to $490 — steep mid-tier","Not a general notification orchestrator (email/SMS)","Inbound event gateway peers (Hookdeck) may fit better for consuming webhooks"],"tags":["webhooks","events","api","delivery"],"scores":{"capabilityFit":9,"pricePerformance":7.8,"latencyReliability":9.2,"developerDx":9,"privacyPosture":8.6,"lockInRisk":4,"maturity":8.6,"overall":8.52,"kind":"curated_estimate","confidence":"high","asOf":"2026-09-14","sources":["Svix pricing: https://www.svix.com/pricing/","Svix: https://www.svix.com"],"notes":"Depot pick for outbound-webhooks. Use Hookdeck when the job is inbound event gateway/debugging; Courier/Knock for multi-channel notify."},"lastReviewed":"2026-09-14","status":"active"},{"id":"hookdeck","slug":"hookdeck","name":"Hookdeck","website":"https://hookdeck.com","tagline":"Inbound event-gateway peer to Svix. Prefer Hookdeck to reliably consume Stripe/GitHub/provider webhooks into…","description":"Inbound event-gateway peer to Svix. Prefer Hookdeck to reliably consume Stripe/GitHub/provider webhooks into agents; Svix to emit webhooks to your customers. Strengths: Best-in-class inbound webhook/event gateway observability; Transformations + issue management for flaky provider webhooks; Clear free → Team ladder for agent backends.","categoryIds":["cat-comms","cat-observability"],"jtbdIds":["outbound-webhooks","event-driven-agent-jobs","infra-observability","inbound-webhook-gateway"],"regions":["global","us"],"pricingNotes":"As of 2026-10-24 (hookdeck.com/pricing, fetched 2026-10-24): Developer 'Always $0 /month' ('Up to 10,000 events', '3-day retention'). Team 'Starts at $39 /month' ('Pay-as-you-go', '7-day retention'). Growth 'Starts at $499 /month' ('Uptime & latency SLAs', '30-day retention'). Enterprise 'Custom'. Extra events: 'Event pricing as low as $0.33 per 100k.'; the rendered text also shows '$1.00 per additional 100k', and the FAQ says 'The usage billing is pro-rated by 10k.' (10,001 events on Team 'is 0.3$'). Volume thresholds are not published in the page text; Depot records this as a conflict and follows the FAQ example as an upper bound. 'Includes 5 events/second of throughput per destination (per project)'. 'The price of an event includes all its retries.' See plan limits (/api/estimate) for the encoded plans.","strengths":["Best-in-class inbound webhook/event gateway observability","Transformations + issue management for flaky provider webhooks","Clear free → Team ladder for agent backends"],"weaknesses":["Outbound customer-facing webhook portal weaker than Svix","Growth SLA tier expensive for indie agents","Not a multi-channel notification product"],"tags":["webhooks","events","gateway","observability"],"scores":{"capabilityFit":8.8,"pricePerformance":8.2,"latencyReliability":9,"developerDx":8.8,"privacyPosture":8.2,"lockInRisk":3.8,"maturity":8.2,"overall":8.42,"kind":"curated_estimate","confidence":"high","asOf":"2026-09-14","sources":["Hookdeck pricing: https://hookdeck.com/pricing","Hookdeck: https://hookdeck.com"],"notes":"Inbound event-gateway peer to Svix. Prefer Hookdeck to reliably consume Stripe/GitHub/provider webhooks into agents; Svix to emit webhooks to your customers."},"lastReviewed":"2026-09-14","status":"active"},{"id":"meilisearch-cloud","slug":"meilisearch-cloud","name":"Meilisearch Cloud","website":"https://www.meilisearch.com","tagline":"Typesense/Algolia peer for product + RAG hybrid search. Prefer Typesense when hourly HA clusters are clearer…","description":"Typesense/Algolia peer for product + RAG hybrid search. Prefer Typesense when hourly HA clusters are clearer; Meilisearch for DX/OSS velocity. Strengths: Delightful typo-tolerant search DX + hybrid/vector; Predictable instance economics vs per-request Algolia; Strong OSS escape hatch.","categoryIds":["cat-search-scrape","cat-vector-memory"],"jtbdIds":["hybrid-search-rag","neural-search","enterprise-search"],"regions":["global","us","eu"],"pricingNotes":"As of 2026-09-14 (meilisearch.com/pricing): Cloud from ~**$20–$30/mo** usage-based or resource-based (example XS 0.5 vCPU/1GB ~$23/mo with disk). 14-day trial. OSS self-host free. Enterprise custom (SSO/SLA).","strengths":["Delightful typo-tolerant search DX + hybrid/vector","Predictable instance economics vs per-request Algolia","Strong OSS escape hatch"],"weaknesses":["Smaller enterprise search feature set than Elastic/Algolia","You still size instances vs fully serverless","AI merchandising features gated enterprise"],"tags":["search","hybrid","oss","vector"],"scores":{"capabilityFit":8.4,"pricePerformance":8.6,"latencyReliability":8.4,"developerDx":9,"privacyPosture":7.8,"lockInRisk":2.8,"maturity":8.2,"overall":8.36,"kind":"curated_estimate","confidence":"high","asOf":"2026-09-14","sources":["Meilisearch pricing: https://www.meilisearch.com/pricing","Meilisearch: https://www.meilisearch.com"],"notes":"Typesense/Algolia peer for product + RAG hybrid search. Prefer Typesense when hourly HA clusters are clearer; Meilisearch for DX/OSS velocity."},"lastReviewed":"2026-09-14","status":"active"},{"id":"rime","slug":"rime","name":"Rime","website":"https://www.rime.ai","tagline":"Realtime TTS latency peer. Depot pick stays Cartesia; ElevenLabs for library; Hume expressive; Rime when Mist…","description":"Realtime TTS latency peer. Depot pick stays Cartesia; ElevenLabs for library; Hume expressive; Rime when Mist TTFA + self-host options win. Strengths: Ultra-low-latency Mist model for voice agents; LiveKit/Pipecat/Twilio integrations; Self-host/VPC path for regulated voice.","categoryIds":["cat-media-gen","cat-inference-apis"],"jtbdIds":["realtime-tts","speech-synthesis","voice-agents"],"regions":["global","us"],"pricingNotes":"As of 2026-09-14 (rime.ai/pricing): Mist v3 **$0.03/1k chars** (~$0.03/min); Coda **$0.05/1k chars**. Starter includes ~800 free minutes / ~3k free minutes on new accounts (marketing copy varies — confirm console). Enterprise volume + VPC/on-prem + BAA. Claimed Mist TTFA P50 ~37ms.","strengths":["Ultra-low-latency Mist model for voice agents","LiveKit/Pipecat/Twilio integrations","Self-host/VPC path for regulated voice"],"weaknesses":["Smaller voice library than ElevenLabs","Free-minute marketing figures inconsistent across page sections — pin console","Less expressive than Hume for emotion"],"tags":["tts","voice","low-latency","agents"],"scores":{"capabilityFit":8.6,"pricePerformance":8.8,"latencyReliability":8.4,"developerDx":8.6,"privacyPosture":8.4,"lockInRisk":4.2,"maturity":7.4,"overall":8.27,"kind":"curated_estimate","confidence":"high","asOf":"2026-09-14","sources":["Rime pricing: https://www.rime.ai/pricing","Rime: https://www.rime.ai"],"notes":"Realtime TTS latency peer. Depot pick stays Cartesia; ElevenLabs for library; Hume expressive; Rime when Mist TTFA + self-host options win."},"lastReviewed":"2026-09-14","status":"active"},{"id":"koyeb","slug":"koyeb","name":"Koyeb","website":"https://www.koyeb.com","tagline":"Serverless GPU/app peer for EU-friendly deploys. Depot neocloud pick stays Nebius; interruptible Salad…","description":"Serverless GPU/app peer for EU-friendly deploys. Depot neocloud pick stays Nebius; interruptible Salad; Python jobs Modal; Koyeb when Git+GPU scale-to-zero DX wins. Strengths: Clear public GPU $/hr + scale-to-zero serverless; EU/US/Asia regions without hyperscaler theater; Git deploy + GPU inference endpoints in one DX.","categoryIds":["cat-gpu-cloud","cat-cpu-compute"],"jtbdIds":["serverless-gpu","neocloud-gpu","app-hosting","predictable-gpu"],"regions":["global","us","eu","ap"],"pricingNotes":"As of 2026-09-14 (koyeb.com/pricing): Per-second serverless. GPUs: RTX-4000 $0.50; L4 $0.70; A6000 $0.75; L40S $1.20; A100 $1.60; A100 SXM $2.15; H100 **$2.50**; H200 $3.00; B200 $5.50 (multi-GPU SKUs listed). Pro $29/mo +compute ($10 included); Scale $299; Enterprise custom. Postgres Small ~$0.04/hr.","strengths":["Clear public GPU $/hr + scale-to-zero serverless","EU/US/Asia regions without hyperscaler theater","Git deploy + GPU inference endpoints in one DX"],"weaknesses":["Not the absolute cheapest neocloud vs Salad/Vast interruptible","Platform fee (Pro $29) on top of compute","Smaller capacity pools than CoreWeave/Nebius"],"tags":["gpu","serverless","neocloud","eu"],"scores":{"capabilityFit":8.2,"pricePerformance":8.4,"latencyReliability":8.2,"developerDx":8.8,"privacyPosture":7.8,"lockInRisk":3.6,"maturity":7.8,"overall":8.12,"kind":"curated_estimate","confidence":"high","asOf":"2026-09-14","sources":["Koyeb pricing: https://www.koyeb.com/pricing","Koyeb: https://www.koyeb.com"],"notes":"Serverless GPU/app peer for EU-friendly deploys. Depot neocloud pick stays Nebius; interruptible Salad; Python jobs Modal; Koyeb when Git+GPU scale-to-zero DX wins."},"lastReviewed":"2026-09-14","status":"active"},{"id":"clickhouse-cloud","slug":"clickhouse-cloud","name":"ClickHouse Cloud","website":"https://clickhouse.com","tagline":"Depot pick for agent-analytics-warehouse (event/log OLAP). MotherDuck remains sql-warehouse-agent-context for…","description":"Depot pick for agent-analytics-warehouse (event/log OLAP). MotherDuck remains sql-warehouse-agent-context for DuckDB-shaped app analytics; Snowflake/Databricks if already committed. Strengths: Best-in-class analytical SQL for high-volume agent logs/events; Compute-storage separation + idle pause on Basic; Strong SQL DX for agent tools querying history.","categoryIds":["cat-vector-memory","cat-observability","cat-cpu-compute"],"jtbdIds":["sql-warehouse-agent-context","agent-analytics-warehouse","cost-monitoring","llm-audit-log-warehouse"],"regions":["global","us","eu"],"pricingNotes":"As of 2026-10-24 (clickhouse.com/pricing, fetched 2026-10-24; list rates shown for 'N. Virginia (us-east-1)'): 'Pay only for the compute and storage you use, metered separately and scaled to zero when idle.' Basic: Storage '$25.30 Per 1TB / mo', Compute '$0.2181 Per unit / hr', 'Up to 1 TB storage', '8-12 GiB total memory'. Scale: Storage '$25.30 Per 1TB / mo', Compute '$0.2985 Per unit / hr', 'Unlimited storage'. Enterprise: Storage '$25.30 Per 1TB / mo', Compute '$0.3903 Per unit / hr'. 'One compute unit is 8 GiB of RAM and 2 vCPUs.' 'Compute is metered per minute and billed per compute-unit-hour at the rate for your plan, cloud provider and region shown above.' 'Storage is metered on the compressed size of your data in object storage, plus backups.' 'Storage rates are the same across plans and vary only by cloud provider and region.' Data transfer: 'public internet egress starting at $0.1152 / GB', 'inter region egress starting at $0.0312 / GB'; 'Ingress is free.' 'ClickPipes pricing depends on the connector type.' 'Start free with $300 in credits for 30 days.' 'Pricing is identical whether you buy through a marketplace or directly from ClickHouse, and self-service customers pay monthly by credit card in USD.' No monthly plan totals are stated in the page text (the calculator needs inputs), so none are quoted.","strengths":["Best-in-class analytical SQL for high-volume agent logs/events","Compute-storage separation + idle pause on Basic","Strong SQL DX for agent tools querying history"],"weaknesses":["Not a OLTP/app DB — wrong default vs Postgres","Always-on Scale bills climb fast","Unit economics need workload modeling (not a flat SaaS seat)"],"tags":["analytics","warehouse","sql","logs","olap"],"scores":{"capabilityFit":9,"pricePerformance":8,"latencyReliability":8.8,"developerDx":8.4,"privacyPosture":8.2,"lockInRisk":4.5,"maturity":9,"overall":8.38,"kind":"curated_estimate","confidence":"high","asOf":"2026-10-19","sources":["ClickHouse Cloud billing overview: https://clickhouse.com/docs/cloud/manage/billing/overview","ClickHouse pricing: https://clickhouse.com/pricing"],"notes":"Depot pick for agent-analytics-warehouse (event/log OLAP). MotherDuck remains sql-warehouse-agent-context for DuckDB-shaped app analytics; Snowflake/Databricks if already committed."},"lastReviewed":"2026-10-19","status":"active"},{"id":"nango","slug":"nango","name":"Nango","website":"https://nango.dev","tagline":"Depot pick for saas-oauth-integrations (embed customer SaaS OAuth/sync). Composio remains agent-tool-router…","description":"Depot pick for saas-oauth-integrations (embed customer SaaS OAuth/sync). Composio remains agent-tool-router for MCP/tool calling breadth. Strengths: OAuth + sync infrastructure for embedding SaaS integrations; Self-host escape hatch vs pure SaaS tool routers; MCP/auth tooling oriented to productized integrations.","categoryIds":["cat-sandboxes-browsers","cat-auth-payments"],"jtbdIds":["saas-oauth-integrations","agent-tool-router"],"regions":["global","us","eu"],"pricingNotes":"As of 2026-09-14 (nango.dev/pricing via live HTML): Free $0 (hard monthly caps). Pay-as-you-go **$50/mo** including $50 credits; usage examples ~$0.29/connection, $0.72/hour sync, $0.5/GB. Growth add-on ~$450/mo. Enterprise custom. Self-host ELv2 option.","strengths":["OAuth + sync infrastructure for embedding SaaS integrations","Self-host escape hatch vs pure SaaS tool routers","MCP/auth tooling oriented to productized integrations"],"weaknesses":["Not a turnkey 1000-tool agent router like Composio","Pricing page sometimes bot/409 flaky — confirm console","You still implement per-API actions"],"tags":["oauth","integrations","sync","mcp","oss"],"scores":{"capabilityFit":8.4,"pricePerformance":8,"latencyReliability":8.2,"developerDx":8.6,"privacyPosture":8.4,"lockInRisk":3.5,"maturity":7.8,"overall":8.14,"kind":"curated_estimate","confidence":"high","asOf":"2026-09-14","sources":["Nango pricing: https://nango.dev/pricing","Nango: https://nango.dev"],"notes":"Depot pick for saas-oauth-integrations (embed customer SaaS OAuth/sync). Composio remains agent-tool-router for MCP/tool calling breadth."},"lastReviewed":"2026-09-14","status":"active"},{"id":"oxylabs","slug":"oxylabs","name":"Oxylabs","website":"https://oxylabs.io","tagline":"Enterprise proxy peer to Bright Data. Prefer Scrapfly/ZenRows/Firecrawl for agent-default scrape; Oxylabs…","description":"Enterprise proxy peer to Bright Data. Prefer Scrapfly/ZenRows/Firecrawl for agent-default scrape; Oxylabs when residential fleet + compliance RFPs demand it. Strengths: Enterprise proxy + scraper API peer to Bright Data; Public starting prices (not pure contact-sales theater); EU-rooted compliance posture.","categoryIds":["cat-search-scrape"],"jtbdIds":["enterprise-proxy","web-scrape","serp-scrape"],"regions":["global","eu","us"],"pricingNotes":"As of 2026-09-14 (oxylabs.io/pricing): Residential proxies from **$6/GB**; Web Scraper API from **$49/mo**; Web Unblocker / mobile / ISP SKUs listed with starting bands. Free trials on several products. ISO 27001 claimed.","strengths":["Enterprise proxy + scraper API peer to Bright Data","Public starting prices (not pure contact-sales theater)","EU-rooted compliance posture"],"weaknesses":["Still expensive vs Scrapfly/ZenRows for light agent scrape","Full GB tiers often sales-assisted at scale","Overkill for LLM-ready crawl (prefer Firecrawl)"],"tags":["proxy","scrape","residential","enterprise"],"scores":{"capabilityFit":8.8,"pricePerformance":7,"latencyReliability":8.8,"developerDx":7.8,"privacyPosture":8,"lockInRisk":4.8,"maturity":8.8,"overall":7.99,"kind":"curated_estimate","confidence":"high","asOf":"2026-09-14","sources":["Oxylabs pricing: https://oxylabs.io/pricing","Oxylabs: https://oxylabs.io"],"notes":"Enterprise proxy peer to Bright Data. Prefer Scrapfly/ZenRows/Firecrawl for agent-default scrape; Oxylabs when residential fleet + compliance RFPs demand it."},"lastReviewed":"2026-09-14","status":"active"},{"id":"elastic-serverless","slug":"elastic-serverless","name":"Elasticsearch Serverless","website":"https://www.elastic.co","tagline":"Enterprise hybrid/vector search when already in Elastic. Prefer Typesense/Meilisearch/Algolia for product…","description":"Enterprise hybrid/vector search when already in Elastic. Prefer Typesense/Meilisearch/Algolia for product search defaults; OpenSearch self-host if you need classic nodes without Elastic licensing theater. Strengths: Search AI Lake + vector/RAG enterprise gravity; Published VCU serverless card (not only classic nodes); Agent Builder / workflows for Elastic-centric agents.","categoryIds":["cat-search-scrape","cat-vector-memory","cat-observability"],"jtbdIds":["hybrid-search-rag","enterprise-search","neural-search","rag-enterprise"],"regions":["global","us","eu"],"pricingNotes":"As of 2026-09-14 (elastic.co/pricing/serverless-search): Ingest from **$0.14**/VCU-hr; Search from **$0.09**; ML from **$0.07**; retention from **$0.047**/GB-mo; egress from $0.05/GB. Managed LLM $4.50/$21 per 1M in/out. Inference Service from $0.08/1M tokens. Agent Builder executions from $0.025 after 1k free (effective 2026-05-01). Illustrative dev ~$24–27/mo; prod 20GB ~$190–210/mo.","strengths":["Search AI Lake + vector/RAG enterprise gravity","Published VCU serverless card (not only classic nodes)","Agent Builder / workflows for Elastic-centric agents"],"weaknesses":["Easy to overspend vs Typesense/Meilisearch for simple apps","Classic Elastic complexity still leaks into ops thinking","Managed LLM rates expensive vs dedicated inference hosts"],"tags":["elasticsearch","serverless","vector","rag","search"],"scores":{"capabilityFit":9,"pricePerformance":6.8,"latencyReliability":9,"developerDx":7.6,"privacyPosture":8.4,"lockInRisk":5.5,"maturity":9.4,"overall":8.06,"kind":"curated_estimate","confidence":"high","asOf":"2026-09-14","sources":["Elasticsearch Serverless pricing: https://www.elastic.co/pricing/serverless-search","Elastic: https://www.elastic.co"],"notes":"Enterprise hybrid/vector search when already in Elastic. Prefer Typesense/Meilisearch/Algolia for product search defaults; OpenSearch self-host if you need classic nodes without Elastic licensing theater."},"lastReviewed":"2026-09-14","status":"active"},{"id":"linkup","slug":"linkup","name":"Linkup","website":"https://www.linkup.so","tagline":"Agent-search runner when you want transparent per-call EU-leaning search. Depot pick stays Exa; Tavily…","description":"Agent-search runner when you want transparent per-call EU-leaning search. Depot pick stays Exa; Tavily co-default; Parallel for deep research tasks. Strengths: Agent-native search/fetch/research with crisp per-call card; EU-friendly posture + SOC2; Good alternative when Exa/Tavily procurement stalls.","categoryIds":["cat-search-scrape"],"jtbdIds":["agent-search","web-grounded-qa","llm-ready-extract"],"regions":["global","eu","us"],"pricingNotes":"As of 2026-09-14 (docs.linkup.so pricing + linkup.so/pricing): Search standard/flash **$0.005** (searchResults) / **$0.006** (sourcedAnswer/structured); deep **$0.05/$0.055**. Fetch $0.001–$0.011 by mode/JS/schema. Research S–XL $0.25–$2.50. ~$20 monthly credit top-up for eligible accounts (~4k standard searches). SOC2; ZDR on Enterprise. x402 USDC option.","strengths":["Agent-native search/fetch/research with crisp per-call card","EU-friendly posture + SOC2","Good alternative when Exa/Tavily procurement stalls"],"weaknesses":["Smaller mindshare than Exa/Tavily/Parallel","Deep/research calls get pricey fast","Credit top-up model needs balance monitoring"],"tags":["search","agents","fetch","research","eu"],"scores":{"capabilityFit":8.4,"pricePerformance":8.6,"latencyReliability":8.2,"developerDx":8.8,"privacyPosture":8.6,"lockInRisk":3.5,"maturity":7.2,"overall":8.23,"kind":"curated_estimate","confidence":"high","asOf":"2026-09-14","sources":["Linkup pricing docs: https://docs.linkup.so/pages/documentation/platform/pricing","Linkup pricing: https://www.linkup.so/pricing"],"notes":"Agent-search runner when you want transparent per-call EU-leaning search. Depot pick stays Exa; Tavily co-default; Parallel for deep research tasks."},"lastReviewed":"2026-09-14","status":"active"},{"id":"kling","slug":"kling","name":"Kling AI","website":"https://kling.ai","tagline":"Video-gen peer to Runway/Luma/Fal. Prefer Fal for multi-model media API DX; Kling when quality demos win and…","description":"Video-gen peer to Runway/Luma/Fal. Prefer Fal for multi-model media API DX; Kling when quality demos win and you accept CN-origin API packaging. Medium until unauthenticated unit card is durable. Strengths: Strong text/image-to-video quality contender; Public API unit economics separate from consumer seats; Useful Fal/Runway peer for video agents.","categoryIds":["cat-media-gen"],"jtbdIds":["video-generation","media-generation"],"regions":["global","cn","us"],"pricingNotes":"As of 2026-09-14: Official API (kling.ai/dev) is prepaid **units** distinct from consumer app subscriptions — secondary/official summaries ~$0.14/unit with video roughly **$0.084–$0.168/sec** (720p–1080p) and higher for 4K; packages from ~$9.80/100 units. Consumer app plans do **not** include API. As of 2026-10-25 https://kling.ai/dev/pricing answers HTTP 200 to plain curl (no redirect, byte-identical twice about 10 s apart) but is client-rendered: the tag-stripped server HTML has 73 characters of text and no price; the unit card sits only in the page's inline script data, which includes 'Video API Pricing 1 Unit = $0.14 (list price)' and 'Image API Pricing 1 Unit = $0.0035 (list price)'. Stored figures are not re-priced here; any price change goes through the reviewed --catalog step.","strengths":["Strong text/image-to-video quality contender","Public API unit economics separate from consumer seats","Useful Fal/Runway peer for video agents"],"weaknesses":["API price card on https://kling.ai/dev/pricing is client-rendered (fetched 2026-10-25, HTTP 200: no price in the 73-character server-rendered text; prices only in inline script data); geo variance not tested","CN-origin ops/compliance considerations","App subscription ≠ API access (common buyer trap)"],"tags":["video","media","api","cn"],"scores":{"capabilityFit":8.8,"pricePerformance":7.8,"latencyReliability":7.8,"developerDx":7.2,"privacyPosture":6.2,"lockInRisk":5,"maturity":7.6,"overall":7.58,"kind":"curated_estimate","confidence":"medium","asOf":"2026-09-14","sources":["Kling AI API pricing: https://kling.ai/dev/pricing","Kling AI: https://kling.ai"],"notes":"Video-gen peer to Runway/Luma/Fal. Prefer Fal for multi-model media API DX; Kling when quality demos win and you accept CN-origin API packaging. Medium until unauthenticated unit card is durable."},"lastReviewed":"2026-09-14","status":"active"},{"id":"astra-db","slug":"astra-db","name":"DataStax Astra DB","website":"https://www.datastax.com/products/datastax-astra","tagline":"Managed vector peer for Cassandra estates / huge write scale. Depot defaults remain Supabase pgvector /…","description":"Managed vector peer for Cassandra estates / huge write scale. Depot defaults remain Supabase pgvector / Qdrant / Pinecone. Medium until RU card is as crisp as Pinecone. Strengths: Cassandra-backed vector + JSON for huge scale; Multi-cloud marketplace distribution; Good when you already standardize on DataStax.","categoryIds":["cat-vector-memory"],"jtbdIds":["managed-vector","rag-memory","billion-scale-vector"],"regions":["global","us","eu"],"pricingNotes":"As of 2026-09-14 (datastax.com Astra/watsonx.data pricing): Pay-per-use from **$1.00 USD per Resource Unit**; Subscription dedicated from **~$83.30/mo**. Cassandra/Astra RU worksheets published (Starter 341 RU/mo → XL annual). Vector/Milvus RU/hr bands (Starter 1.25 → Large 16.50 RU/hr). Now packaged under IBM watsonx.data lakehouse marketing — still the Astra DB serverless/vector path for Cassandra estates. Pin console RU export before large RAG commits.","strengths":["Cassandra-backed vector + JSON for huge scale","Multi-cloud marketplace distribution","Good when you already standardize on DataStax"],"weaknesses":["Public card thinner than Pinecone/Qdrant/TurboPuffer","RU mental model less intuitive for agent builders","Overkill vs pgvector for most app RAG"],"tags":["vector","cassandra","serverless","rag"],"scores":{"capabilityFit":8.4,"pricePerformance":7.4,"latencyReliability":8.8,"developerDx":7.6,"privacyPosture":8.2,"lockInRisk":5.2,"maturity":8.8,"overall":7.93,"kind":"curated_estimate","confidence":"high","asOf":"2026-09-14","sources":["Astra DB product/pricing: https://www.datastax.com/products/datastax-astra/pricing","DataStax Astra: https://www.datastax.com/products/datastax-astra"],"notes":"Managed vector peer for Cassandra estates / huge write scale. Depot defaults remain Supabase pgvector / Qdrant / Pinecone. Medium until RU card is as crisp as Pinecone."},"lastReviewed":"2026-09-14","status":"active"},{"id":"posthog","slug":"posthog","name":"PostHog","website":"https://posthog.com","tagline":"Flags/analytics peer. Depot oss-feature-flags pick stays GrowthBook; LaunchDarkly for AgentControl; PostHog…","description":"Flags/analytics peer. Depot oss-feature-flags pick stays GrowthBook; LaunchDarkly for AgentControl; PostHog when product analytics+flags should share one bill. Strengths: OSS + product analytics + flags + experiments in one bill; EU hosting option; Good agent-product telemetry without Datadog tax.","categoryIds":["cat-observability","cat-auth-payments"],"jtbdIds":["oss-feature-flags","progressive-agent-rollout","experiment-tracking"],"regions":["global","us","eu"],"pricingNotes":"As of 2026-09-14 (posthog.com/pricing): Generous free product quotas (e.g. 1M analytics events / flags / error tracking bands — exact free units on calculator). Usage-based per product with billing limits; EU cloud (Frankfurt) available. Session replay, LLM analytics, workflows metered separately.","strengths":["OSS + product analytics + flags + experiments in one bill","EU hosting option","Good agent-product telemetry without Datadog tax"],"weaknesses":["Not a specialist LLM tracer (Langfuse/Helicone deeper)","Many metered products — easy to enable bill surprises","Flags depth vs LaunchDarkly enterprise controls"],"tags":["analytics","flags","oss","experiments","eu"],"scores":{"capabilityFit":8.4,"pricePerformance":8.8,"latencyReliability":8.4,"developerDx":9,"privacyPosture":8.4,"lockInRisk":4,"maturity":8.6,"overall":8.41,"kind":"curated_estimate","confidence":"high","asOf":"2026-09-14","sources":["PostHog pricing: https://www.posthog.com/pricing","PostHog: https://posthog.com"],"notes":"Flags/analytics peer. Depot oss-feature-flags pick stays GrowthBook; LaunchDarkly for AgentControl; PostHog when product analytics+flags should share one bill."},"lastReviewed":"2026-09-14","status":"active"},{"id":"sentry","slug":"sentry","name":"Sentry","website":"https://sentry.io","tagline":"Depot pick for agent-error-observability (exceptions/traces in the app hosting the agent). Keep…","description":"Depot pick for agent-error-observability (exceptions/traces in the app hosting the agent). Keep Langfuse/Helicone for LLM request tracing; Datadog when enterprise APM is already standardized. Strengths: Default error/trace monitoring every agent backend needs; MCP access on free plan for agentized debugging; Mature SDKs across languages/runtimes.","categoryIds":["cat-observability"],"jtbdIds":["agent-error-observability","infra-observability"],"regions":["global","us","eu"],"pricingNotes":"As of 2026-09-14 (sentry.io/pricing): Developer **Free** (1 user, 5k errors class quotas + MCP access). Team from **$26/mo**; Business **$80/mo** (annual prepaid defaults). PAYG overages for errors/spans/replays/profiling. Seer AI debugging agent add-on.","strengths":["Default error/trace monitoring every agent backend needs","MCP access on free plan for agentized debugging","Mature SDKs across languages/runtimes"],"weaknesses":["Not a substitute for LLM-specific eval/tracing (Langfuse/Braintrust)","Span/replay overages can surprise","AI Seer features are extra"],"tags":["errors","tracing","apm","mcp","observability"],"scores":{"capabilityFit":8.8,"pricePerformance":8.4,"latencyReliability":9,"developerDx":9,"privacyPosture":8,"lockInRisk":4.2,"maturity":9.4,"overall":8.56,"kind":"curated_estimate","confidence":"high","asOf":"2026-09-14","sources":["Sentry pricing: https://sentry.io/pricing/","Sentry: https://sentry.io"],"notes":"Depot pick for agent-error-observability (exceptions/traces in the app hosting the agent). Keep Langfuse/Helicone for LLM request tracing; Datadog when enterprise APM is already standardized."},"lastReviewed":"2026-09-14","status":"active"},{"id":"cloudflare-browser-rendering","slug":"cloudflare-browser-rendering","name":"Cloudflare Browser Rendering","website":"https://developers.cloudflare.com/browser-rendering/","tagline":"Edge browser peer when already on Cloudflare. Depot cloud-browser pick stays Browserbase; Anchor/Hyperbrowser…","description":"Edge browser peer when already on Cloudflare. Depot cloud-browser pick stays Browserbase; Anchor/Hyperbrowser for computer-use. Medium until Workers Paid browser-hour card re-exported. Strengths: Edge-colocated browsers beside Workers/Durable Objects; Same Cloudflare bill as R2/Workers AI/AI Gateway; Good for light crawl/screenshot agents.","categoryIds":["cat-sandboxes-browsers"],"jtbdIds":["cloud-browser","web-scrape","self-host-browser"],"regions":["global"],"pricingNotes":"As of 2026-10-24 (developers.cloudflare.com/browser-run/pricing, redirected from browser-rendering/platform/pricing, fetched 2026-10-24; the product is now named 'Cloudflare Browser Run'; 'Last updated Apr 21, 2026'): 'Available on Free and Paid plans'. Quick Actions: 'Charged for browser hours only.' Browser Sessions (Puppeteer, Playwright, CDP): 'Direct browser control, charged for both browser hours and concurrent browsers.' 'Browser hours are shared across all methods.' Workers Free: browser hours '10 minutes per day', concurrent browsers '3 browsers'. Workers Paid: '10 hours per month, then $0.09 per additional hour'; concurrent browsers '10 browsers' ('averaged monthly'), 'then $2.00 per additional browser'. 'Cloudflare calculates concurrent browsers as the monthly average of your daily peak usage.' Monthly browser hours are 'rounded to the nearest whole hour'. 'If a Quick Actions request fails with a waitForTimeout error, the browser session is not charged.'","strengths":["Edge-colocated browsers beside Workers/Durable Objects","Same Cloudflare bill as R2/Workers AI/AI Gateway","Good for light crawl/screenshot agents"],"weaknesses":["Not a stealth/computer-use fleet like Browserbase/Anchor","Quotas tied to Workers plan complexity","Less session DX than Browserbase for long agent loops"],"tags":["browser","cloudflare","workers","edge"],"scores":{"capabilityFit":7.8,"pricePerformance":8.8,"latencyReliability":8.6,"developerDx":8.2,"privacyPosture":8.4,"lockInRisk":5,"maturity":8,"overall":8.04,"kind":"curated_estimate","confidence":"high","asOf":"2026-09-14","sources":["Cloudflare Browser Rendering: https://developers.cloudflare.com/browser-rendering/","Cloudflare Browser Rendering pricing: https://developers.cloudflare.com/browser-rendering/platform/pricing/"],"notes":"Edge browser peer when already on Cloudflare. Depot cloud-browser pick stays Browserbase; Anchor/Hyperbrowser for computer-use. Medium until Workers Paid browser-hour card re-exported."},"lastReviewed":"2026-09-14","status":"active"}]}