Free resources for agents

Downloadable Qwen3 0.6B base weights. This row is this checkpoint only, not every Qwen model.

Limit not stated

License: Apache-2.0

Jobs: open-model-inference, budget-open-inference

LICENSE file fetched 2026-10-03 is Apache License 2.0. Weights are the model repo, not this license URL. LICENSE re-fetched twice 2026-10-25 (HTTP 200, byte-identical): 'Apache License Version 2.0, January 2004'.

https://huggingface.co/Qwen/Qwen3-0.6B-Base/raw/main/LICENSE · as of 2026-10-25

Downloadable DeepSeek-V3.2 weights from the Hugging Face repo whose LICENSE file is MIT.

Limit not stated

License: MIT

Jobs: open-model-inference, budget-open-inference

This is DeepSeek-V3.2, not DeepSeek-V3. V3 used a separate LICENSE-MODEL and is not indexed here. LICENSE re-fetched twice 2026-10-25 (HTTP 200, byte-identical): 'MIT License'.

https://huggingface.co/deepseek-ai/DeepSeek-V3.2/raw/main/LICENSE · as of 2026-10-25

Whisper code and model weights in the OpenAI GitHub repository.

Limit not stated

License: MIT

Jobs: speech-to-text, production-stt

LICENSE fetched 2026-10-03 is MIT, copyright 2022 OpenAI. Re-fetched twice 2026-10-25: the GitHub blob page (HTTP 200) changes between fetches (page chrome), so the raw file https://raw.githubusercontent.com/openai/whisper/main/LICENSE was also fetched twice (HTTP 200, byte-identical): 'MIT License' / 'Copyright (c) 2022 OpenAI'.

https://github.com/openai/whisper/blob/main/LICENSE · as of 2026-10-25

Gemma models listed in the Appendix of the Gemma Terms of Use page (for example Gemma 1, 1.1, 2, 3 and 3n, FunctionGemma, EmbeddingGemma, PaliGemma, ShieldGemma, CodeGemma, RecurrentGemma, T5Gemma, VaultGemma and TranslateGemma), usable under that license with its Section 3.2 use restrictions. Gemma 4 is not covered by these terms.

Limit not stated

License: Gemma Terms of Use

Provider: google-ai

Jobs: open-model-inference, wasm-edge-small-models

Terms page fetched twice 2026-10-24 (identical text; 'Last modified: April 1, 2026', footer 'Last updated 2026-04-01 UTC'): 'The terms below apply to Gemma models listed in the Appendix at bottom of this page.' and 'For Gemma 4 terms, see the Gemma 4 license'. That link points to https://ai.google.dev/gemma/apache_2, a page titled 'Apache License 2.0' (fetched twice 2026-10-24); Gemma 4 is not indexed in this row. Section 3.2: 'You must not use any of the Gemma Services: for the restricted uses set forth in the Gemma Prohibited Use Policy'. The page now also says 'This documentation is deprecated. Please visit the new documentation at gemma.ai'; the terms are still served at this URL (HTTP 200, no redirect). Download path is https://ai.google.dev/gemma/docs/get_started: 'You can download official Google Gemma model variants and community-created variants from Kaggle Models and Hugging Face'. Re-checked 2026-10-26 (v0.12.131): stored quotes still match. https://ai.google.dev/gemma/terms, https://ai.google.dev/gemma/apache_2 and https://ai.google.dev/gemma/docs/get_started were each fetched twice (~10 s apart, plain curl, no login, following redirects): HTTP 200, no redirect, HTML differs only inside inline scripts, tag-stripped text identical; the terms page still reads 'Last modified: April 1, 2026' and 'Last updated 2026-04-01 UTC'. Every quote above string-matched the whitespace-normalized text; license and values unchanged (Depot data decision to re-date only, approvalRef 'Ethan delegation 2026-10-21 19:41 ET t55u', decidedBy Agent Depo).

https://ai.google.dev/gemma/terms · as of 2026-10-26

Downloadable Qwen2.5-Coder 7B Instruct weights. This checkpoint only.

Limit not stated

License: Apache-2.0

Jobs: coding-agent, open-model-inference

LICENSE file fetched 2026-10-03 is Apache License 2.0. LICENSE re-fetched twice 2026-10-25 (HTTP 200, byte-identical): 'Apache License Version 2.0, January 2004'.

https://huggingface.co/Qwen/Qwen2.5-Coder-7B-Instruct/raw/main/LICENSE · as of 2026-10-25

Downloadable Qwen3-8B weights.

Limit not stated

License: Apache-2.0

Jobs: open-model-inference, budget-open-inference

LICENSE file fetched 2026-10-03 is Apache License 2.0. Not a claim about every Qwen3 checkpoint. LICENSE re-fetched twice 2026-10-25 (HTTP 200, byte-identical): 'Apache License Version 2.0, January 2004'.

https://huggingface.co/Qwen/Qwen3-8B/raw/main/LICENSE · as of 2026-10-25

Downloadable Microsoft Phi-4 weights.

Limit not stated

License: MIT

Jobs: open-model-inference, budget-open-inference

LICENSE file fetched 2026-10-03 is the MIT License. LICENSE re-fetched twice 2026-10-25 (HTTP 200, byte-identical): 'MIT License'.

https://huggingface.co/microsoft/phi-4/raw/main/LICENSE · as of 2026-10-25

Downloadable OpenAI gpt-oss-20b weights.

Limit not stated

License: Apache-2.0

Jobs: open-model-inference, budget-open-inference

LICENSE file fetched 2026-10-03 is Apache License 2.0. LICENSE re-fetched twice 2026-10-25 (HTTP 200, byte-identical): 'Apache License Version 2.0, January 2004'.

https://huggingface.co/openai/gpt-oss-20b/raw/main/LICENSE · as of 2026-10-25

Downloadable DeepSeek-R1 weights.

Limit not stated

License: MIT

Jobs: open-model-inference

LICENSE file fetched 2026-10-03 is the MIT License. This is DeepSeek-R1, not DeepSeek-V3. LICENSE re-fetched twice 2026-10-25 (HTTP 200, byte-identical): 'MIT License'.

https://huggingface.co/deepseek-ai/DeepSeek-R1/raw/main/LICENSE · as of 2026-10-25

Downloadable SmolLM2 1.7B Instruct weights.

Limit not stated

License: Apache-2.0

Jobs: open-model-inference, budget-open-inference, on-device-edge-batch

Model card front matter fetched 2026-10-03 says license: apache-2.0. A separate LICENSE file was not in the repo (404). README re-fetched twice 2026-10-25 (HTTP 200, byte-identical): front matter 'license: apache-2.0'; raw/main/LICENSE still 404 (twice).

https://huggingface.co/HuggingFaceTB/SmolLM2-1.7B-Instruct/raw/main/README.md · as of 2026-10-25

Downloadable Allen Institute OLMo-2 7B Instruct weights.

Limit not stated

License: Apache-2.0

Jobs: open-model-inference

Model card front matter fetched 2026-10-03 says license: apache-2.0. A separate LICENSE file was 404. README re-fetched twice 2026-10-25 (HTTP 200, byte-identical): front matter 'license: apache-2.0'; raw/main/LICENSE still 404 (twice).

https://huggingface.co/allenai/OLMo-2-1124-7B-Instruct/raw/main/README.md · as of 2026-10-25

Downloadable Mistral-7B-Instruct-v0.3 weights.

Limit not stated

License: Apache-2.0

Jobs: open-model-inference, budget-open-inference

Model card front matter fetched 2026-10-03 says license: apache-2.0. A separate LICENSE file was 404. README re-fetched twice 2026-10-25 (HTTP 200, byte-identical): front matter 'license: apache-2.0'; raw/main/LICENSE still 404 (twice).

https://huggingface.co/mistralai/Mistral-7B-Instruct-v0.3/raw/main/README.md · as of 2026-10-25

Downloadable Qwen2.5-VL 7B Instruct weights for image-text tasks.

Limit not stated

License: Apache-2.0

Jobs: multimodal, open-model-inference

Model card front matter fetched 2026-10-03 says license: apache-2.0. The 3B checkpoint uses a different Qwen Research LICENSE and is not this row. README re-fetched twice 2026-10-25 (HTTP 200, byte-identical): front matter 'license: apache-2.0'; raw/main/LICENSE 404 (twice). The 3B README fetched the same day says 'license_name: qwen-research'.

https://huggingface.co/Qwen/Qwen2.5-VL-7B-Instruct/raw/main/README.md · as of 2026-10-25

Downloadable Microsoft Florence-2-base weights for vision and OCR-style tasks.

Limit not stated

License: MIT

Jobs: layout-aware-ocr, document-parse, multimodal

LICENSE file fetched 2026-10-03 is the MIT License. LICENSE re-fetched twice 2026-10-25 (HTTP 200, byte-identical): 'MIT License'.

https://huggingface.co/microsoft/Florence-2-base/raw/main/LICENSE · as of 2026-10-25

Downloadable BAAI BGE-M3 embedding weights.

Limit not stated

License: MIT

Jobs: embeddings, batch-cheap-embeddings

Model card front matter fetched 2026-10-03 says license: mit. A separate LICENSE file was 404. README re-fetched twice 2026-10-25 (HTTP 200, byte-identical): front matter 'license: mit'; raw/main/LICENSE still 404 (twice).

https://huggingface.co/BAAI/bge-m3/raw/main/README.md · as of 2026-10-25

Downloadable BAAI BGE reranker v2 m3 weights.

Limit not stated

License: Apache-2.0

Jobs: rag-rerank

Model card front matter fetched 2026-10-03 says license: apache-2.0. A separate LICENSE file was 404. README re-fetched twice 2026-10-25 (HTTP 200, byte-identical): front matter 'license: apache-2.0'; raw/main/LICENSE still 404 (twice).

https://huggingface.co/BAAI/bge-reranker-v2-m3/raw/main/README.md · as of 2026-10-25

Downloadable Qwen3 embedding 0.6B weights.

Limit not stated

License: Apache-2.0

Jobs: embeddings, batch-cheap-embeddings

Model card front matter fetched 2026-10-03 says license: apache-2.0. A separate LICENSE file was 404. README re-fetched twice 2026-10-25 (HTTP 200, byte-identical): front matter 'license: apache-2.0'; raw/main/LICENSE still 404 (twice).

https://huggingface.co/Qwen/Qwen3-Embedding-0.6B/raw/main/README.md · as of 2026-10-25

Downloadable Qwen3 reranker 0.6B weights.

Limit not stated

License: Apache-2.0

Jobs: rag-rerank

Model card front matter fetched 2026-10-03 says license: apache-2.0. A separate LICENSE file was 404. README re-fetched twice 2026-10-25 (HTTP 200, byte-identical): front matter 'license: apache-2.0'; raw/main/LICENSE still 404 (twice).

https://huggingface.co/Qwen/Qwen3-Reranker-0.6B/raw/main/README.md · as of 2026-10-25

Downloadable Kokoro 82M text-to-speech weights.

Limit not stated

License: Apache-2.0

Jobs: speech-synthesis, realtime-tts

Model card fetched 2026-10-03 says license: apache-2.0 and Apache-licensed weights. The repo LICENSE file path returned 404; the kokoro code LICENSE on GitHub was also fetched and is Apache-2.0. README re-fetched twice 2026-10-25 (HTTP 200, byte-identical): front matter 'license: apache-2.0'; raw/main/LICENSE still 404 (twice); https://raw.githubusercontent.com/hexgrad/kokoro/main/LICENSE (HTTP 200) is 'Apache License Version 2.0, January 2004'.

https://huggingface.co/hexgrad/Kokoro-82M/raw/main/README.md · as of 2026-10-25

Black Forest Labs FLUX.1-schnell weights. The model card license field says apache-2.0.

Limit not stated

License: Apache-2.0

Jobs: image-generation, text-in-image-generation

cardData.license fetched 2026-10-03 is apache-2.0. gated is auto. LICENSE.md returned 401, so the license text was not copied. Not FLUX.1-dev. Model API re-fetched twice 2026-10-25 (HTTP 200, byte-identical): cardData.license 'apache-2.0', gated 'auto'; LICENSE.md still 401.

https://huggingface.co/api/models/black-forest-labs/FLUX.1-schnell · as of 2026-10-25

Downloadable SDXL base 1.0 weights under the CreativeML Open RAIL++-M license, not Apache.

Limit not stated

License: CreativeML Open RAIL++-M

Jobs: image-generation

LICENSE.md fetched 2026-10-03 is CreativeML Open RAIL++-M dated July 26, 2023. Not labeled Apache. LICENSE.md re-fetched twice 2026-10-25 (HTTP 200, byte-identical): 'CreativeML Open RAIL++-M License dated July 26, 2023'.

https://huggingface.co/stabilityai/stable-diffusion-xl-base-1.0/raw/main/LICENSE.md · as of 2026-10-25

Downloadable GLM-4-9B-Chat weights under the glm-4-9b license. Academic use is stated as free. Commercial use requires registration.

Limit not stated

License: glm-4-9b License

Jobs: open-model-inference

LICENSE fetched 2026-10-03 grants royalty-free academic research and says commercial use requires registration at open.bigmodel.cn, after which commercial use is described as free. Not Apache. Military and illegal uses are prohibited in the same text. Re-fetched twice 2026-10-25: https://huggingface.co/THUDM/glm-4-9b-chat/raw/main/LICENSE now answers HTTP 307 to /zai-org/glm-4-9b-chat/raw/main/LICENSE (HTTP 200, byte-identical), so sourceUrl moved to the redirect target. Text still says 'The glm-4-9b License', 'For users who wish to use the models for commercial purposes, please do so [here](https://open.bigmodel.cn/mla/form)' and 'Registered users are free to use this model for commercial activities'.

https://huggingface.co/zai-org/glm-4-9b-chat/raw/main/LICENSE · as of 2026-10-25

Downloadable jina-embeddings-v3 weights. The model card license is CC-BY-NC-4.0, which is non-commercial.

Limit not stated

License: CC-BY-NC-4.0

Jobs: embeddings

Model card front matter fetched 2026-10-03 says license: cc-by-nc-4.0. Not Apache and not a commercial grant. README re-fetched twice 2026-10-25 (HTTP 200, byte-identical): front matter 'license: cc-by-nc-4.0'; raw/main/LICENSE 404 (twice).

https://huggingface.co/jinaai/jina-embeddings-v3/raw/main/README.md · as of 2026-10-25

Downloadable IBM Granite 3.1 8B Instruct weights.

Limit not stated

License: Apache-2.0

Jobs: open-model-inference

Model card fetched 2026-10-03 says license: apache-2.0 and links the Apache 2.0 text. A separate LICENSE file was 404. README re-fetched twice 2026-10-25 (HTTP 200, byte-identical): front matter 'license: apache-2.0' and '- **License:** [Apache 2.0](https://www.apache.org/licenses/LICENSE-2.0)'; raw/main/LICENSE still 404 (twice).

https://huggingface.co/ibm-granite/granite-3.1-8b-instruct/raw/main/README.md · as of 2026-10-25

Local model runner and OpenAI-compatible server. You still supply weights and your own hardware.

Limit not stated

License: MIT

Jobs: open-model-inference, budget-open-inference

Re-fetched twice 2026-10-25 (~10 s apart, plain curl, no login): the GitHub blob page answered HTTP 200 both times (page chrome differs between fetches), so the raw file https://raw.githubusercontent.com/ollama/ollama/main/LICENSE was also fetched twice (HTTP 200, byte-identical): 'MIT License' / 'Copyright (c) Ollama'. License unchanged.

https://github.com/ollama/ollama/blob/main/LICENSE · as of 2026-10-25

C/C++ GGUF inference CLI and server. Weights are not included.

Limit not stated

License: MIT

Jobs: open-model-inference, budget-open-inference, edge-llm-inference

LICENSE fetched 2026-10-03, copyright 2023-2026 The ggml authors. Re-fetched twice 2026-10-25 (~10 s apart, plain curl, no login): the GitHub blob page answered HTTP 200 both times (page chrome differs between fetches), so the raw file https://raw.githubusercontent.com/ggml-org/llama.cpp/master/LICENSE was also fetched twice (HTTP 200, byte-identical): 'MIT License' / 'Copyright (c) 2023-2026 The ggml authors'. License unchanged.

https://github.com/ggml-org/llama.cpp/blob/master/LICENSE · as of 2026-10-25

Self-hosted GPU inference server with an OpenAI-compatible API. GPU time is your own.

Limit not stated

License: Apache-2.0

Jobs: production-inference-fleet, open-model-inference

Re-fetched twice 2026-10-25 (~10 s apart, plain curl, no login): the GitHub blob page answered HTTP 200 both times (page chrome differs between fetches), so the raw file https://raw.githubusercontent.com/vllm-project/vllm/main/LICENSE was also fetched twice (HTTP 200, byte-identical): 'Apache License Version 2.0, January 2004'. License unchanged.

https://github.com/vllm-project/vllm/blob/main/LICENSE · as of 2026-10-25

Local Whisper inference in C/C++. Weights are separate.

Limit not stated

License: MIT

Jobs: speech-to-text, production-stt

Re-fetched twice 2026-10-25 (~10 s apart, plain curl, no login): the GitHub blob page answered HTTP 200 both times (page chrome differs between fetches), so the raw file https://raw.githubusercontent.com/ggml-org/whisper.cpp/master/LICENSE was also fetched twice (HTTP 200, byte-identical): 'MIT License' / 'Copyright (c) 2023-2026 The ggml authors'. License unchanged.

https://github.com/ggml-org/whisper.cpp/blob/master/LICENSE · as of 2026-10-25

Apple's open-source array framework for on-device model inference.

Limit not stated

License: MIT

Jobs: on-device-edge-batch, open-model-inference

LICENSE fetched 2026-10-03 is MIT, copyright 2023 Apple Inc. Re-fetched twice 2026-10-25 (~10 s apart, plain curl, no login): HTTP 200, byte-identical: 'MIT License' / 'Copyright © 2023 Apple Inc.'. License unchanged.

https://raw.githubusercontent.com/ml-explore/mlx/main/LICENSE · as of 2026-10-25

Open-source serving runtime for LLMs.

Limit not stated

License: Apache-2.0

Jobs: open-model-inference, budget-open-inference

LICENSE fetched 2026-10-03 is Apache License 2.0. Re-fetched twice 2026-10-25 (~10 s apart, plain curl, no login): HTTP 200, byte-identical: 'Apache License Version 2.0, January 2004'. License unchanged.

https://raw.githubusercontent.com/sgl-project/sglang/main/LICENSE · as of 2026-10-25

Mozilla's open-source single-file local LLM runtime.

Limit not stated

License: Apache-2.0

Jobs: open-model-inference, on-device-edge-batch

LICENSE fetched 2026-10-03 is Apache License 2.0, copyright 2023 Mozilla Foundation. Re-fetched twice 2026-10-25 (~10 s apart, plain curl, no login): HTTP 200, byte-identical: 'The Apache 2.0 License' / 'Copyright 2023 Mozilla Foundation'. License unchanged.

https://raw.githubusercontent.com/Mozilla-Ocho/llamafile/main/LICENSE · as of 2026-10-25

Open-source runtime for running models across your own machines.

Limit not stated

License: Apache-2.0

Jobs: open-model-inference, on-device-edge-batch

LICENSE fetched 2026-10-03 is Apache License 2.0. Re-fetched twice 2026-10-25 (~10 s apart, plain curl, no login): HTTP 200, byte-identical: 'Apache License Version 2.0, January 2004'. License unchanged.

https://raw.githubusercontent.com/exo-explore/exo/main/LICENSE · as of 2026-10-25

Organization-level Free Plan rate limits for the chat and classifier model ids in the page payload key freeRows. Not the Developer plan table.

{"scope":"organization","models":{"openai/gpt-oss-120b":{"rpm":30,"rpd":1000,"tpm":8000,"tpd":200000},"openai/gpt-oss-20b":{"rpm":30,"rpd":1000,"tpm":8000,"tpd":200000},"openai/gpt-oss-safeguard-20b":{"rpm":3,"rpd":1000,"tpm":2000,"tpd":200000},"qwen/qwen3.8-27b":{"rpm":30,"rpd":1000,"tpm":8000,"tpd":200000},"meta-llama/llama-prompt-guard-2-22m":{"rpm":30,"rpd":14400,"tpm":15000,"tpd":500000},"meta-llama/llama-prompt-guard-2-86m":{"rpm":30,"rpd":14400,"tpm":15000,"tpd":500000}}}

Provider: groq

Jobs: cheap-volume-llm, ultra-low-latency, open-model-inference

Rate-limits page fetched twice 2026-10-24 about 10 s apart (identical HTML; the page shows no last-updated date). Parsed from the HTML payload key freeRows (the 'Free Plan Limits' tab), which is separate from devRows (Developer plan, not stored). K suffixes expanded (1K=1000, 2K=2000, 14.4K=14400). The page says: 'The following is a high level summary and there may be exceptions to these limits.' and 'You can view the current, exact rate limits for your organization on the limits page in your account settings.' 'Rate limits apply at the organization level, not individual users.' 'Cached tokens do not count towards your rate limits.' The page also says 'some organizations are also subject to separate per-minute limits on input tokens (ITPM) and output tokens (OTPM)'; no ITPM/OTPM values are published, so none are stored. Change on 2026-10-24: freeRows now reads 'openai/gpt-oss-safeguard-20b 3 1K 2K 200K' (RPM 3, TPM 2K); before 0.12.98 this row stored RPM 30 and TPM 8000 for it. The other five model ids and their values are unchanged. No Llama chat model is in freeRows.

https://console.groq.com/docs/rate-limits · as of 2026-10-24

10,000 Neurons per day at no charge on Workers Free and Workers Paid. Workers Free cannot buy more. The pool is shared across models.

{"neuronsPerDay":10000,"resets":"00:00 UTC","sharedAcrossModels":true}

Provider: cloudflare-workers-ai

Jobs: cheap-volume-llm, edge-llm-inference, budget-open-inference

Fetched twice 2026-10-25 about 10 s apart without logging in (HTTP 200, no redirect, identical HTML; the page shows 'Last updated Oct 20, 2026'). 'Our free allocation allows anyone to use a total of 10,000 Neurons per day at no charge.' 'To use more than 10,000 Neurons per day, you need to sign up for the Workers Paid plan.' Free allocation table: 'Workers Free 10,000 Neurons per day N/A - Upgrade to Workers Paid' and 'Workers Paid 10,000 Neurons per day $0.011 / 1,000 Neurons'. 'All limits reset daily at 00:00 UTC.' Paid usage above the allocation is a price row, not a free cap. 'Some models require a paid billing method.' The page lists @cf/moonshotai/kimi-k2.6, @cf/moonshotai/kimi-k2.7-code, @cf/zai-org/glm-5.2, @cf/zai-org/glm-5.3, @cf/zai-org/glm-5.3-flash, @cf/deepseek-ai/deepseek-v4-flash-0731, @cf/deepseek-ai/deepseek-v4.1-flash-1017 and @cf/deepseek-ai/deepseek-v4-pro-0813 ('You can access these models with either the Workers Paid plan or prepaid AI Gateway credits'); these are not free. Per-model token prices are not a $0 SKU.

https://developers.cloudflare.com/workers-ai/platform/pricing/ · as of 2026-10-25

Standard input and output marked Free of charge for models whose free-tier cell says that. Includes gemini-3.8-flash, gemini-3.6-flash, gemini-3.5-flash-lite, gemini-3.1-flash-lite, gemini-3-flash-preview, gemini-2.5-pro, gemini-2.5-flash, and gemini-2.5-flash-lite.

Limit not stated

Provider: google-ai

Jobs: cheap-volume-llm, coding-agent, frontier-chat

Pricing page fetched twice 2026-10-24 (identical text; page footer 'Last updated 2026-10-22 UTC'): the Free tier card lists 'Limited access to certain models' and 'Free input & output tokens'; each listed model's Standard Free Tier input and output cells read 'Free of charge'. gemini-3.7-flash and gemini-3.5-flash, listed in this row before 0.12.96, no longer appear on the page, so they were removed rather than kept unsourced. 'Gemini 4 Argon is not available on the Free tier.' and gemini-3.1-pro-preview standard free-tier cells read 'Not available'. Batch free-tier cells read 'Not available' for several of these models (gemini-3.8-flash, gemini-3-flash-preview, gemini-2.5-pro, gemini-2.5-flash). Image-generation free-tier cells fetched were Not available. The Free tier card says 'Content used to improve our products'. https://ai.google.dev/gemini-api/docs/rate-limits (fetched twice 2026-10-24, identical text, 'Last updated 2026-10-15 UTC') publishes no free-tier RPM, TPM or RPD and says 'View your active rate limits in AI Studio', so limits stay null. Not a $0 price-sort row. The page describes gemini-3.8-flash as 'engineered for long-horizon software engineering' and gemini-2.5-pro as 'A Pro model which excels at coding and complex reasoning tasks.' Re-checked 2026-10-26 (v0.12.131): stored quotes still match. https://ai.google.dev/gemini-api/docs/pricing and https://ai.google.dev/gemini-api/docs/rate-limits were each fetched twice (~10 s apart, plain curl, no login, following redirects): HTTP 200, no redirect, HTML differs only inside inline scripts, tag-stripped text identical; both footers now read 'Last updated 2026-10-26 UTC'. Every quote above string-matched the whitespace-normalized text except the older footer stamps, which are fetch-date stamps, not row values; the eight listed models' Standard free-tier input cells still read 'Free of charge', and gemini-3.7-flash and gemini-3.5-flash are still absent; values unchanged (Depot data decision to re-date only, approvalRef 'Ethan delegation 2026-10-21 19:41 ET t55u', decidedBy Agent Depo).

https://ai.google.dev/gemini-api/docs/pricing · as of 2026-10-26

Model ids ending in :free, and the openrouter/free router, are priced at zero on OpenRouter. Which models have a free entry changes.

{"appliesTo":"model variants with an ID ending in :free","requestsPerMinute":20,"requestsPerDayCreditsPurchasedLessThan10":50,"requestsPerDayCreditsPurchasedAtLeast10":1000,"dailyWindow":"UTC day"}

Provider: openrouter

Jobs: cheap-volume-llm, multi-model-router, budget-open-inference

Fetched twice 2026-10-25 about 10 s apart without logging in: https://openrouter.ai/docs/api/reference/limits answers HTTP 308 to https://openrouter.ai/docs/api_reference/limits (HTTP 200; same tag-stripped text both times; no last-modified header; page metadata dateModified 2026-10-10T16:11:54Z). The HTML table cells are empty until scripts run (the page defines the values as constants), so the numbers are taken from the page's markdown view https://openrouter.ai/docs/api_reference/limits.md (HTTP 200, fetched twice, identical): 'Free usage limits: If you're using a free model variant (with an ID ending in `:free`), the following limits apply:' table 'Credits purchased (all time) | Requests per minute | Requests per day', 'Less than 10 | 20 | 50', 'At least 10 | 20 | 1000'. 'Free-model requests allowed per UTC day (see the free usage limits below)'. 'On free variants, purchase at least 10 credits to raise your daily limit, or switch to the paid variant of the model, which has no platform-level request cap.' 'If your account has a negative credit balance, you may see errors, including for free models.' (the HTML renders the status code as a component: 402 Payment Required). The one-credit-below-threshold grace is shown only as the unrendered placeholder '{FREE_MODEL_CREDITS_THRESHOLD - 1}', so it is not stored. The page does not say whether openrouter/free draws on the same caps. Third-party tables were not copied. Do not copy OpenRouter prices or limits onto other Depot providers.

https://openrouter.ai/docs/api_reference/limits · as of 2026-10-25

Free mode lets you create API keys and use included monthly usage within the limits shown on the Limits page (Admin Panel > API > Limits). The docs publish no numeric free allowance; Pay-as-you-go extends usage beyond the included monthly usage from prepaid credits.

Limit not stated

Provider: mistral

Jobs: cheap-volume-llm, eu-sovereign-llm

The old sourceUrl https://docs.mistral.ai/admin/user-management-finops/tier now returns HTTP 308 to https://docs.mistral.ai/admin/billing-usage/usage-limits (page title 'Usage and limits'), which carries the same Free mode statement; sourceUrl updated to the final URL. Page fetched twice 2026-10-24 about 10 s apart (HTTP 200, identical HTML; the page shows no last-updated date); its markdown view (usage-limits.md) was also fetched twice (identical). Quotes below match the HTML text (link markup removed). 'Your Mistral plan defines the included monthly usage and limits available to your Organization.' 'Free mode lets you create API keys and use included monthly usage within the limits shown on the Limits page.' 'Pay-as-you-go lets you extend usage beyond included monthly usage, and this extra usage draws from your prepaid credits. It isn't a separate plan for the API.' 'API rate limits define how much traffic your Organization can send.' 'If the Organization reaches its monthly spending limit, API access can be suspended until the next month begins or an admin increases the limit.' limits stays null: the page publishes no numeric free cap, rate limit or included-usage amount (it only says completion limits are listed per model as tokens per minute and requests per second on the account Limits page). The Ambassador credits and 'free for a limited amount of time' wording noted on 2026-10-03 no longer appears on this page.

https://docs.mistral.ai/admin/billing-usage/usage-limits · as of 2026-10-24

Trial (evaluation) key for the Chat models on the page, 20 req/min each. Free but limited in usage, not a production key.

{"apiCallsPerMonth":1000,"requestsPerMinute":20,"monthlyCapSharedAcrossTrialKey":true}

Provider: cohere

Jobs: frontier-chat

Page fetched twice 2026-10-25 about 10 s apart without logging in (HTTP 200, no redirect, identical HTML; response header 'last-modified: Fri, 23 Oct 2026 21:27:48 GMT'; the page itself shows no last-updated date). 'Cohere offers two kinds of API keys: evaluation keys (free but limited in usage), and production keys (paid and much less limited in usage).' 'Trial keys (and prod keys on newer Chat model variants) are limited to 1,000 API calls a month.' 'Prod keys work like trial keys for newer model variants such as Command A Reasoning.' Chat API (per model) table: every listed model has a trial rate limit of 20 req / min (Command A+, Command A Reasoning, Command A Translate, Command A Vision, Command A, Command R+, Command R, Command R7B, North Mini Code), e.g. 'Command A 20 req / min 500 req / min'. The 1,000 calls a month is the trial key's cap, not 1,000 per product. Values unchanged from 2026-10-03.

https://docs.cohere.com/docs/rate-limits · as of 2026-10-25

$5 in free credits after adding a verified payment method; they expire 30 days after they're granted and can be used across all public models (Shared Inference). When they run out or expire, API and Playground access stop until you purchase credits.

{"creditsUsd":5,"expiresDaysAfterGrant":30,"requiresVerifiedPaymentMethod":true,"models":{"gpt-oss-120b":{"rpm":5,"uncachedTpm":30000,"totalTpm":90000,"tph":1000000,"tpd":1000000},"qwen-3.8-27b":{"rpm":5,"uncachedTpm":30000,"totalTpm":90000,"tph":1000000,"tpd":1000000}}}

Provider: cerebras

Jobs: ultra-low-latency, cheap-volume-llm

Rate-limits page fetched twice 2026-10-24 about 10 s apart (identical HTML; the page shows no last-updated date); the page's markdown view (rate-limits.md) was also fetched twice (identical) and its Free Trial table has the same values. Quotes below match the HTML text (curly apostrophes shown straight). Free Trial table (HTML text): 'gpt-oss-120b 5 30K 90K 1M 1M' and 'qwen-3.8-27b 1 5 30K 90K 1M 1M' (the second '1' is the image-limit footnote marker; columns RPM, Uncached TPM, Total TPM, TPH, TPD); K/M suffixes expanded (30K=30000, 90K=90000, 1M=1000000). The table lists exactly these two models. qwen-3.8-27b footnote: 'Image limits: 2 per request, 10 MiB total request payload'. The page says: 'New accounts receive $5 in free credits after adding a verified payment method. These credits expire 30 days after they're granted and can be used across all public models.' 'If you skip adding a payment method at sign-up, Playground and API access remain inactive until you do.' 'Is there a permanently free tier? No.' 'Cerebras doesn't currently offer a no-cost tier that renews automatically or a per-model always-free allowance.' 'This provides an overview of general limits, though specific cases may vary.' 'Rate limits apply at the organization level, not the user level, and vary based on the model.' Developer (Pay as You Go) values are not stored on this free row (a paid tier); since v0.12.129 they are in the plan-limits table (data/limits.json, cerebras / Developer (Pay as You Go)). All stored values re-confirmed unchanged on 2026-10-24. Not a Depot price-sort input.

https://inference-docs.cerebras.ai/support/rate-limits · as of 2026-10-24

Embedding models on Workers AI, including bge-m3 and qwen3-embedding-0.6b, draw from the same 10,000 neurons/day. No separate free token cap is stated.

{"sharedNeuronsPerDay":10000,"separateTokenCap":null}

Provider: cloudflare-workers-ai

Jobs: embeddings, batch-cheap-embeddings

Fetched twice 2026-10-25 about 10 s apart without logging in (HTTP 200, no redirect, identical HTML; the page shows 'Last updated Oct 20, 2026'). 'Our free allocation allows anyone to use a total of 10,000 Neurons per day at no charge.' The 'Embeddings model pricing' table lists @cf/baai/bge-small-en-v1.5, bge-base-en-v1.5, bge-large-en-v1.5, bge-m3, @cf/pfnet/plamo-embedding-1b and @cf/qwen/qwen3-embedding-0.6b, e.g. '@cf/baai/bge-m3 $0.012 per M input tokens 1075 neurons per M input tokens'. Those are the paid unit prices billed in neurons above the allocation, not a free token cap; the page states no separate free embedding cap.

https://developers.cloudflare.com/workers-ai/platform/pricing/ · as of 2026-10-25

gemini-embedding-2 standard text, image, audio, and video input marked Free of charge.

Limit not stated

Provider: google-ai

Jobs: embeddings, batch-cheap-embeddings

Pricing page fetched twice 2026-10-24 (identical text; page footer 'Last updated 2026-10-22 UTC'): gemini-embedding-2 is 'Our first multimodal embedding model, mapping text, images, video, audio, and PDFs into a unified embedding space.' Its Standard Free Tier cells for Text, Image, Audio and Video input price read 'Free of charge'; the Batch free-tier cells read 'Not available'. No numeric free-tier quota is on the pricing page. https://ai.google.dev/gemini-api/docs/rate-limits (fetched twice 2026-10-24, identical text, 'Last updated 2026-10-15 UTC') publishes no free-tier RPM, TPM or RPD and says 'View your active rate limits in AI Studio', so limits stay null. Re-checked 2026-10-26 (v0.12.131): stored quotes still match. https://ai.google.dev/gemini-api/docs/pricing and https://ai.google.dev/gemini-api/docs/rate-limits were each fetched twice (~10 s apart, plain curl, no login, following redirects): HTTP 200, no redirect, HTML differs only inside inline scripts, tag-stripped text identical; both footers now read 'Last updated 2026-10-26 UTC'. Every quote above string-matched the whitespace-normalized text except the older footer stamps, which are fetch-date stamps, not row values; values unchanged (Depot data decision to re-date only, approvalRef 'Ethan delegation 2026-10-21 19:41 ET t55u', decidedBy Agent Depo).

https://ai.google.dev/gemini-api/docs/pricing · as of 2026-10-26

Trial-key Embed rate limits. Still inside the trial key's 1,000 API calls a month.

{"apiCallsPerMonth":1000,"embedInputsPerMinute":2000,"embedImageInputsPerMinute":5,"monthlyCapSharedAcrossTrialKey":true}

Provider: cohere

Jobs: embeddings

Page fetched twice 2026-10-25 about 10 s apart without logging in (HTTP 200, no redirect, identical HTML; response header 'last-modified: Fri, 23 Oct 2026 21:27:48 GMT'; the page itself shows no last-updated date). 'Trial keys (and prod keys on newer Chat model variants) are limited to 1,000 API calls a month.' Other API Endpoints table (Trial rate limit, then Production rate limit): 'Embed 2,000 inputs / min 2,000 inputs / min'; 'Embed (Images) 5 inputs / min 400 inputs / min'. The production Embed text limit is the same 2,000 inputs/min; the monthly 1,000-call cap is the trial-key constraint. Not a forever unlimited embed tier. Values unchanged from 2026-10-03.

https://docs.cohere.com/docs/rate-limits · as of 2026-10-25

Embedding API with a free API key at the free-key rate limit.

{"requestsPerMinute":100,"tokensPerMinute":100000,"ipRequestsPer60Seconds":10000}

Provider: jina

Jobs: embeddings

Page fetched twice 2026-10-25 about 10 s apart without logging in (HTTP 200, no redirect, identical HTML; response header 'last-modified: Thu, 22 Oct 2026 01:48:53 GMT'; the page itself shows no last-updated date). FAQ: 'Rate limits depend on your API key type: Free: 100 RPM, 100K TPM'. Rate table, Embedding API row, w/ Free API Key column: '100 RPM & 100,000 TPM' (the w/o API Key column shows a block icon). FAQ: 'There is an additional IP-based limit of 10,000 requests per 60 seconds to prevent abuse.' FAQ: 'New users get an auto-generated API key with free tokens usable across any of our models.' and 'No, our API keys do not have an expiration date.' The page states no free token amount, so no token balance is stored (a 10 million token grant is not on the page). Paid and Premium columns are not the free tier and were not copied. Values unchanged from 2026-10-03.

https://jina.ai/embeddings/ · as of 2026-10-25

Free Plan speech-to-text limits for whisper-large-v3 and whisper-large-v3-turbo.

{"scope":"organization","models":{"whisper-large-v3":{"rpm":20,"rpd":2000,"audioSecondsPerHour":7200,"audioSecondsPerDay":28800},"whisper-large-v3-turbo":{"rpm":20,"rpd":2000,"audioSecondsPerHour":7200,"audioSecondsPerDay":28800}}}

Provider: groq

Jobs: speech-to-text, production-stt

Rate-limits page fetched twice 2026-10-24 about 10 s apart (identical HTML; the page shows no last-updated date). Same freeRows payload: whisper-large-v3 and whisper-large-v3-turbo both read RPM 20, RPD 2K, ASH 7.2K, ASD 28.8K (unchanged). ASH/ASD kept as audio seconds ('ASH: Audio seconds per hour', 'ASD: Audio seconds per day'), not tokens; TPM/TPD are '-' on the page and are not stored. Same exception and account-page caveat as the chat row. Not a separate quota from the org's other Free Plan limits.

https://console.groq.com/docs/rate-limits · as of 2026-10-24

Free Plan text-to-speech limits for two Orpheus voices.

{"scope":"organization","models":{"canopylabs/orpheus-arabic-saudi":{"rpm":10,"rpd":100,"tpm":1200,"tpd":3600},"canopylabs/orpheus-v1-english":{"rpm":10,"rpd":100,"tpm":1200,"tpd":3600}}}

Provider: groq

Jobs: speech-synthesis, realtime-tts

Rate-limits page fetched twice 2026-10-24 about 10 s apart (identical HTML; the page shows no last-updated date). Same freeRows payload: canopylabs/orpheus-arabic-saudi and canopylabs/orpheus-v1-english both read RPM 10, RPD 100, TPM 1.2K, TPD 3.6K (unchanged). Same exception and account-page caveat as the chat row. Developer-plan numbers were not copied.

https://console.groq.com/docs/rate-limits · as of 2026-10-24

Whisper and Whisper large v3 turbo speech-to-text on the shared 10,000 neurons/day free allocation. No separate free audio-minute cap is stated.

{"sharedNeuronsPerDay":10000,"separateAudioCap":null}

Provider: cloudflare-workers-ai

Jobs: speech-to-text

The page prices @cf/openai/whisper and @cf/openai/whisper-large-v3-turbo at $0.0005 per audio minute in neurons. That is the paid unit, not an always-free minute cap. Re-checked 2026-10-19: speech-synthesis tag removed from this row (Whisper is STT); TTS models on the same page are a separate row, cloudflare-workers-ai-tts.

https://developers.cloudflare.com/workers-ai/platform/pricing/ · as of 2026-10-19

@cf/myshell-ai/melotts and @cf/deepgram/aura-1 text-to-speech on the same shared 10,000 neurons/day free allocation as every Workers AI model. No separate free TTS cap is stated.

{"sharedNeuronsPerDay":10000,"separateAudioCap":null}

Provider: cloudflare-workers-ai

Jobs: speech-synthesis

The audio pricing table (fetched 2026-10-19) lists @cf/myshell-ai/melotts at $0.0002 per audio minute and @cf/deepgram/aura-1 at $0.015 per 1k characters input, billed in neurons. Those are paid units, not free quotas; the free part is the shared daily neuron allocation.

https://developers.cloudflare.com/workers-ai/platform/pricing/ · as of 2026-10-19

gemini-3.5-transcribe standard audio input and text output marked Free of charge. gemini-3.5-transcribe-live (real-time streaming transcription over WebSockets) standard cells are also Free of charge.

Limit not stated

Provider: google-ai

Jobs: speech-to-text, multilingual-stt

Pricing page fetched twice 2026-10-26 (identical text; page footer 'Last updated 2026-10-26 UTC'): gemini-3.5-transcribe is 'Our speech-to-text model with automatic language detection, speaker diarization, word-level timestamps, and custom vocabulary biasing.' Its Standard Free Tier input and output cells read 'Free of charge'. gemini-3.5-transcribe-live is 'Our low-latency, real-time speech-to-text model for bidirectional streaming audio transcription over WebSockets.' and its Standard Free Tier input and output cells also read 'Free of charge'. No numeric audio quota. The page's per-minute paid estimates ('Estimated pricing is based on 25 audio tokens per second for input and 175 text tokens per minute for output', with blended rates of ~$0.005 per min for Transcribe and ~$0.009 per min for Live Transcribe) are paid-tier estimates, not a free cap. https://ai.google.dev/gemini-api/docs/rate-limits (fetched twice 2026-10-26, identical text, 'Last updated 2026-10-26 UTC') publishes no free-tier RPM, TPM or RPD and says 'View your active rate limits in AI Studio', so limits stay null. Re-quoted in v0.12.130 (previously fetched twice 2026-10-24, page footer 'Last updated 2026-10-22 UTC'); gemini-3.5-transcribe-live added to this row in v0.12.130 as a Depot data decision (approvalRef 'Ethan delegation 2026-10-21 19:41 ET t55u', decidedBy Agent Depo); no price stored, not a $0 price-sort row.

https://ai.google.dev/gemini-api/docs/pricing · as of 2026-10-26

gemini-3.8-live, gemini-3.8-live-extended-thinking and gemini-3.1-flash-live-preview (low-latency audio-to-audio models for real-time voice agents and live dialogue) standard input and output marked Free of charge.

Limit not stated

Provider: google-ai

Jobs: voice-agents

Pricing page fetched twice 2026-10-26 (identical text; page footer 'Last updated 2026-10-26 UTC'): the section 'Gemini 3.8 Live, Gemini 3.8 Live Extended Thinking, and Gemini 3.1 Flash Live Preview' ('gemini-3.8-live, gemini-3.8-live-extended-thinking, and gemini-3.1-flash-live-preview') reads 'Our low-latency, audio-to-audio models optimized for real-time voice agents and live dialogue, including background reasoning support with Extended Thinking.' Its one table covers all three models: 'Standard Free Tier Paid Tier, per 1M tokens in USD Input price Free of charge' and 'Output price (including thinking tokens) Free of charge'. 'Used to improve our products Yes No' (Yes on the free tier). No numeric free audio cap; paid-tier per-token and per-minute prices are not stored. https://ai.google.dev/gemini-api/docs/rate-limits (fetched twice 2026-10-26, identical text, 'Last updated 2026-10-26 UTC') publishes no free-tier RPM, TPM or RPD for these models and says 'View your active rate limits in AI Studio', so limits stay null. Not included: gemini-3.5-live-translate-preview ('Our low-latency, real-time speech to speech translation model that supports 70+ languages.'; its Standard free-tier cells also read 'Free of charge'), because the catalog has no speech-translation job and it is not a voice-agent model; gemini-2.5-flash-native-audio-preview-12-2025 ('Our Live API native audio models optimized for higher quality audio outputs with better pacing, voice naturalness, verbosity, and mood.'), because its table header reads 'Free Tier Paid Tier, per 1M tokens in USD' with no 'Standard' label, so a Standard free tier was not string-matched; gemini-robotics-er-2-preview and -streaming-preview (no matching job). Added in v0.12.132 as a Depot data decision (approvalRef 'Ethan delegation 2026-10-21 19:41 ET t55u', decidedBy Agent Depo); no price stored, not a $0 price-sort row. Update 2026-10-26 (v0.12.134): gemini-2.5-flash-native-audio-preview-12-2025 is now indexed in its own row, gemini-2.5-flash-native-audio-free-tier, which quotes its unlabeled 'Free Tier' cells without calling them standard; this row is unchanged.

https://ai.google.dev/gemini-api/docs/pricing · as of 2026-10-26

gemini-2.5-flash-native-audio-preview-12-2025 (Live API native audio model optimized for higher quality audio outputs) free-tier input and output marked Free of charge.

Limit not stated

Provider: google-ai

Jobs: voice-agents

Pricing page fetched twice 2026-10-26 (~10 s apart, plain curl, no login, following redirects; HTTP 200, no redirect, identical tag-stripped text; page footer 'Last updated 2026-10-26 UTC'): the section 'Gemini 2.5 Flash Native Audio (Live API)' ('gemini-2.5-flash-native-audio-preview-12-2025') reads 'Our Live API native audio models optimized for higher quality audio outputs with better pacing, voice naturalness, verbosity, and mood.' The section prints one table with no Standard, Batch, Flex or Priority label: 'Free Tier Paid Tier, per 1M tokens in USD Input price Free of charge' and 'Output price (including thinking tokens) Free of charge'. 'Used to improve our products Yes No' (Yes on the free tier). Data call left open in v0.12.132: because the header carries no 'Standard' label, this row does not call the cells standard; it records only that the section's one free-tier column reads 'Free of charge' for input and output (cost free_tier_unspecified). No numeric free audio cap; paid-tier per-token prices are not stored. https://ai.google.dev/gemini-api/docs/rate-limits (fetched twice 2026-10-26, identical text, 'Last updated 2026-10-26 UTC') does not name this model, publishes no free-tier RPM, TPM or RPD for it and says 'View your active rate limits in AI Studio', so limits stay null. Kept as its own row, not folded into gemini-live-free-tier, because the page prints it as a separate section and table. Added in v0.12.134 as a Depot data decision (approvalRef 'Ethan delegation 2026-10-21 19:41 ET t55u', decidedBy Agent Depo); no price stored, not a $0 price-sort row.

https://ai.google.dev/gemini-api/docs/pricing · as of 2026-10-26

gemini-3.8-flash-tts standard text input and audio output marked Free of charge. gemini-3.8-flash-lite-tts, gemini-3.1-flash-tts-preview and gemini-2.5-flash-preview-tts standard cells are also Free of charge.

Limit not stated

Provider: google-ai

Jobs: speech-synthesis, realtime-tts

Pricing page fetched twice 2026-10-26 (identical text; page footer 'Last updated 2026-10-26 UTC'): gemini-3.8-flash-tts is 'Our 3.8 Flash Text-to-Speech audio model engineered for studio-grade voice fidelity, expressive acting, and long-form stability.' Its Standard Free Tier input and output cells read 'Free of charge', as do those of gemini-3.8-flash-lite-tts, gemini-3.1-flash-tts-preview and gemini-2.5-flash-preview-tts ('Our 2.5 Flash text-to-speech audio model optimized for price-performant, low-latency, controllable speech generation.'). Batch TTS free tier is 'Not available' for 3.8 Flash TTS and 2.5 Flash Preview TTS. gemini-2.5-pro-preview-tts is not included: its Standard free-tier cells read 'Not available'. No numeric free audio cap. https://ai.google.dev/gemini-api/docs/rate-limits (fetched twice 2026-10-26, identical text, 'Last updated 2026-10-26 UTC') publishes no free-tier RPM, TPM or RPD and says 'View your active rate limits in AI Studio', so limits stay null. Re-quoted in v0.12.130 (previously fetched twice 2026-10-24, page footer 'Last updated 2026-10-22 UTC'); gemini-2.5-flash-preview-tts added to this row in v0.12.130 as a Depot data decision (approvalRef 'Ethan delegation 2026-10-21 19:41 ET t55u', decidedBy Agent Depo); no price stored, not a $0 price-sort row.

https://ai.google.dev/gemini-api/docs/pricing · as of 2026-10-26

Open-source local text-to-speech engine.

Limit not stated

License: MIT

Jobs: speech-synthesis, realtime-tts

LICENSE.md fetched 2026-10-03 is MIT, copyright 2022 Michael Hansen. Voice packs are a separate download; rhasspy/piper-voices card says license mit. Re-fetched twice 2026-10-25 (~10 s apart, plain curl, no login): HTTP 200, no redirect, byte-identical: 'MIT License' / 'Copyright (c) 2022 Michael Hansen'. License unchanged.

https://raw.githubusercontent.com/rhasspy/piper/master/LICENSE.md · as of 2026-10-25

Open-source local speech synthesizer.

Limit not stated

License: GPL-3.0

Jobs: speech-synthesis

COPYING fetched 2026-10-03 is GNU GPL version 3. Re-fetched twice 2026-10-25 (~10 s apart, plain curl, no login): HTTP 200, no redirect, byte-identical: 'GNU GENERAL PUBLIC LICENSE Version 3, 29 June 2007'. License unchanged.

https://raw.githubusercontent.com/espeak-ng/espeak-ng/master/COPYING · as of 2026-10-25

@cf/baai/bge-reranker-base draws from the same 10,000 neurons/day.

{"sharedNeuronsPerDay":10000,"separateTokenCap":null}

Provider: cloudflare-workers-ai

Jobs: rag-rerank

Fetched twice 2026-10-25 about 10 s apart without logging in (HTTP 200, no redirect, identical HTML; the page shows 'Last updated Oct 20, 2026'). 'Our free allocation allows anyone to use a total of 10,000 Neurons per day at no charge.' '@cf/baai/bge-reranker-base $0.003 per M input tokens 283 neurons per M input tokens' is the paid unit billed in neurons, not a free token cap; the page states no separate free rerank cap.

https://developers.cloudflare.com/workers-ai/platform/pricing/ · as of 2026-10-25

Trial-key Rerank limit of 10 requests per minute, inside the 1,000 calls/month trial cap.

{"apiCallsPerMonth":1000,"requestsPerMinute":10,"monthlyCapSharedAcrossTrialKey":true}

Provider: cohere

Jobs: rag-rerank

Page fetched twice 2026-10-25 about 10 s apart without logging in (HTTP 200, no redirect, identical HTML; response header 'last-modified: Fri, 23 Oct 2026 21:27:48 GMT'; the page itself shows no last-updated date). 'Trial keys (and prod keys on newer Chat model variants) are limited to 1,000 API calls a month.' Other API Endpoints table (Trial rate limit, then Production rate limit): 'Rerank 10 req / min 1,000 req / min'. Production rerank (1,000 req/min) is not the free tier. Values unchanged from 2026-10-03.

https://docs.cohere.com/docs/rate-limits · as of 2026-10-25

Reranker calls on a free Jina API key at the free-key rate limit shown on the embeddings page.

{"requestsPerMinute":100,"tokensPerMinute":100000,"ipRequestsPer60Seconds":10000}

Provider: jina

Jobs: rag-rerank

Page fetched twice 2026-10-25 about 10 s apart without logging in (HTTP 200, no redirect, identical HTML; response header 'last-modified: Thu, 22 Oct 2026 01:48:53 GMT'; the page itself shows no last-updated date). Rate table, Reranker API row, w/ Free API Key column: '100 RPM & 100,000 TPM' (the w/o API Key column shows a block icon), matching the FAQ line 'Rate limits apply per API key: Free: 100 RPM, 100K TPM'. FAQ: 'There is an additional IP-based limit of 10,000 requests per 60 seconds to prevent abuse.' FAQ: 'New users get an auto-generated API key with free tokens usable across any of our models.' and 'No, our API keys do not have an expiration date.' The page states no free token amount, so no token balance is stored (a 10 million token grant is not on the page). Paid and Premium columns are not the free tier and were not copied. Values unchanged from 2026-10-03.

https://jina.ai/embeddings/ · as of 2026-10-25

Open-source local reranking library.

Limit not stated

License: Apache-2.0

Jobs: rag-rerank

LICENSE fetched 2026-10-03 is Apache License 2.0. Re-fetched twice 2026-10-25 (~10 s apart, plain curl, no login): HTTP 200, no redirect, byte-identical: 'Apache License Version 2.0, January 2004'. License unchanged.

https://raw.githubusercontent.com/PrithivirajDamodaran/FlashRank/main/LICENSE · as of 2026-10-25

Image models on the pricing page, including FLUX.1 schnell, draw from the same 10,000 neurons/day. No separate free image count is stated.

{"sharedNeuronsPerDay":10000,"separateImageCap":null}

Provider: cloudflare-workers-ai

Jobs: image-generation, text-in-image-generation

Fetched twice 2026-10-25 about 10 s apart without logging in (HTTP 200, no redirect, identical HTML; the page shows 'Last updated Oct 20, 2026'). 'Our free allocation allows anyone to use a total of 10,000 Neurons per day at no charge.' The 'Image model pricing' table lists FLUX.1 schnell ('@cf/black-forest-labs/flux-1-schnell $0.0000528 per 512x512 tile $0.0001056 per step'), lucid-origin, phoenix-1.0, flux-2-dev, flux-2-klein-4b and flux-2-klein-9b. Tile, step and megapixel prices are paid units billed in neurons, not a free image allowance; the page states no separate free image count.

https://developers.cloudflare.com/workers-ai/platform/pricing/ · as of 2026-10-25

Open-source node UI for running image models locally.

Limit not stated

License: GPL-3.0

Jobs: image-generation

LICENSE fetched 2026-10-03 is GNU GPL version 3. Model weights you load keep their own licenses. Re-fetched twice 2026-10-25 (~10 s apart, plain curl, no login): HTTP 200, no redirect, byte-identical: 'GNU GENERAL PUBLIC LICENSE Version 3, 29 June 2007'. License unchanged.

https://raw.githubusercontent.com/comfyanonymous/ComfyUI/master/LICENSE · as of 2026-10-25

Open-source local runtime for Stable Diffusion family models.

Limit not stated

License: MIT

Jobs: image-generation

LICENSE fetched 2026-10-03 is MIT. Weights are not included in this license. Re-fetched twice 2026-10-25 (~10 s apart, plain curl, no login): HTTP 200, no redirect, byte-identical: 'MIT License' / 'Copyright (c) 2023 leejet'. License unchanged.

https://raw.githubusercontent.com/leejet/stable-diffusion.cpp/master/LICENSE · as of 2026-10-25

Grounding with Google Search is free of charge up to 500 requests per day on Gemini 2.5 Flash, shared with Flash-Lite.

{"requestsPerDay":500,"sharedWith":"gemini-2.5-flash and gemini-2.5-flash-lite"}

Provider: google-ai

Jobs: web-grounded-qa, agent-search

Pricing page fetched twice 2026-10-24 (identical text; page footer 'Last updated 2026-10-22 UTC'): the gemini-2.5-flash free-tier Grounding with Google Search cell reads 'Free of charge, up to 500 RPD (limit shared with Flash-Lite RPD)' and the gemini-2.5-flash-lite cell reads 'Free of charge, up to 500 RPD (limit shared with Flash RPD)'. Not available for Pro on the free tier (gemini-2.5-pro cell 'Not available'). Gemini 3.x and newer tables mark Grounding with Google Search as Not available on the free tier, with a separate paid allowance of '5,000 free search requests per month (shared across all Gemini 3 and newer models), then $14 per 1,000 requests.' Those were not copied onto 2.5. Maps grounding also lists 500 RPD for 2.5 Flash and is not indexed as Search. Re-checked 2026-10-26 (v0.12.131): stored quotes still match. https://ai.google.dev/gemini-api/docs/pricing was each fetched twice (~10 s apart, plain curl, no login, following redirects): HTTP 200, no redirect, HTML differs only inside inline scripts, tag-stripped text identical; its footer now reads 'Last updated 2026-10-26 UTC'. Every quote above string-matched the whitespace-normalized text except the older footer stamp, which is a fetch-date stamp, not a row value; limits (500 RPD shared between 2.5 Flash and Flash-Lite) and other values unchanged (Depot data decision to re-date only, approvalRef 'Ethan delegation 2026-10-21 19:41 ET t55u', decidedBy Agent Depo).

https://ai.google.dev/gemini-api/docs/pricing · as of 2026-10-26

Using Google AI Studio, Google's hosted studio for trying Gemini models, is free of charge in all available regions. This is the studio, not a published API quota; Gemini API calls follow the Free and Paid tiers on the same page.

Limit not stated

Provider: google-ai

Jobs: frontier-chat

Pricing page fetched twice 2026-10-24 (identical text; page footer 'Last updated 2026-10-22 UTC'): 'Google AI Studio usage is free of charge in all available regions.' The Free tier card lists 'Google AI Studio access', and model cards link 'Try it in Google AI Studio'. No numeric studio quota or rate limit is stated, so limits stay null. Reclassified free_docs -> free_tool in 0.12.94 because the quote is about using a hosted product, not reading docs. https://aistudio.google.com is client-rendered (no usable text) and is not a source. Re-checked 2026-10-26 (v0.12.131): stored quotes still match. https://ai.google.dev/gemini-api/docs/pricing was each fetched twice (~10 s apart, plain curl, no login, following redirects): HTTP 200, no redirect, HTML differs only inside inline scripts, tag-stripped text identical; its footer now reads 'Last updated 2026-10-26 UTC'. Every quote above string-matched the whitespace-normalized text except the older footer stamp, which is a fetch-date stamp, not a row value; values unchanged (Depot data decision to re-date only, approvalRef 'Ethan delegation 2026-10-21 19:41 ET t55u', decidedBy Agent Depo).

https://ai.google.dev/gemini-api/docs/pricing · as of 2026-10-26

Browser automation library for Chromium, Firefox, and WebKit. The library is free; hosted browsers are not included.

Limit not stated

License: Apache-2.0

Provider: playwright

Jobs: self-host-browser, cloud-browser, computer-use-browser

Re-fetched twice 2026-10-25 (~10 s apart, plain curl, no login): GitHub blob page HTTP 200, no redirect (page chrome differs between fetches); raw LICENSE fetched twice too, HTTP 200, byte-identical: 'Apache License Version 2.0, January 2004' / 'Portions Copyright (c) Microsoft Corporation.'. License unchanged.

https://github.com/microsoft/playwright/blob/main/LICENSE · as of 2026-10-25

Python library that drives a browser from an agent. Model calls and browsers are not included.

Limit not stated

License: MIT

Jobs: web-agent, computer-use-browser, browser-agent-loop

Re-fetched twice 2026-10-25 (~10 s apart, plain curl, no login): GitHub blob page HTTP 200, no redirect (page chrome differs between fetches); raw LICENSE fetched twice too, HTTP 200, byte-identical: 'MIT License' / 'Copyright (c) 2024 Gregor Zunic'. License unchanged.

https://github.com/browser-use/browser-use/blob/main/LICENSE · as of 2026-10-25

Local eval CLI and library. No hosted quota is included.

Limit not stated

License: MIT

Provider: promptfoo

Jobs: oss-evals, evals, adversarial-red-team-evals, continuous-eval-ci

LICENSE fetched 2026-10-03 is MIT. Copyright line says Promptfoo 2025. README fetched 2026-10-03 says to automate checks in CI/CD. It does not say the check blocks a merge, and it does not name GitHub or GitLab. README fetched 2026-10-19 says promptfoo is a CLI and library for evaluating and red-teaming LLM-based apps, with red teaming and vulnerability scanning. Re-fetched twice 2026-10-25 (~10 s apart, plain curl, no login): GitHub blob page HTTP 200, no redirect (page chrome differs between fetches); raw LICENSE fetched twice too, HTTP 200, byte-identical: 'Copyright (c) Promptfoo 2025' / 'Permission is hereby granted, free of charge'. The file is the MIT permission text without an 'MIT License' heading. License unchanged.

https://github.com/promptfoo/promptfoo/blob/main/LICENSE · as of 2026-10-25

UK AISI eval framework. Model provider calls are billed by those providers.

Limit not stated

License: MIT

Jobs: oss-evals, evals

Re-fetched twice 2026-10-25 (~10 s apart, plain curl, no login): GitHub blob page HTTP 200, no redirect (page chrome differs between fetches); raw LICENSE fetched twice too, HTTP 200, byte-identical: 'MIT License' / 'Copyright (c) 2024 UK AI Security Institute'. License unchanged.

https://github.com/UKGovernmentBEIS/inspect_ai/blob/main/LICENSE · as of 2026-10-25

Self-hosted LLM tracing. Code outside the ee/ directories is MIT. Hosted Langfuse Cloud is not this row.

Limit not stated

License: MIT Expat outside ee/

Provider: langfuse

Jobs: llm-tracing, agent-trace-session-replay

ee/, web/src/ee/, worker/src/ee/, and packages/shared/src/server/ee/ use the license in ee/LICENSE. Not indexed as a free cloud tier. Re-fetched twice 2026-10-25 (~10 s apart, plain curl, no login): GitHub blob page HTTP 200, no redirect (page chrome differs between fetches); raw LICENSE fetched twice too, HTTP 200, byte-identical: 'Copyright (c) 2023-2026 ClickHouse, Inc.' / '"MIT Expat" license'. Note corrected 2026-10-25 (v0.12.120): raw LICENSE (https://raw.githubusercontent.com/langfuse/langfuse/main/LICENSE) re-fetched twice (~10 s apart, plain curl, no login), HTTP 200, no redirect, byte-identical; it lists '"ee/", "web/src/ee/", "worker/src/ee/", and/or "packages/shared/src/server/ee/"' as licensed under ee/LICENSE (GitHub commits API: LICENSE changed 2026-10-23, 'chore(license): include packages/shared/src/server/ee in the enterprise license (#17877)') and content outside them under the '"MIT Expat" license'. License value 'MIT Expat outside ee/' unchanged. Flag resolved.

https://github.com/langfuse/langfuse/blob/main/LICENSE · as of 2026-10-25

Self-hosted OpenAI-compatible gateway. Code outside enterprise/ is MIT. Provider tokens are not included.

Limit not stated

License: MIT outside enterprise/

Jobs: llm-ai-gateway, multi-model-router

enterprise/ uses enterprise/LICENSE. No free hosted quota is stated here. Re-fetched twice 2026-10-25 (~10 s apart, plain curl, no login): GitHub blob page HTTP 200, no redirect (page chrome differs between fetches); raw LICENSE fetched twice too, HTTP 200, byte-identical: 'available under the MIT license' / 'Copyright (c) 2023 Berri AI'. The LICENSE still names only the 'enterprise/' directory. License unchanged.

https://github.com/BerriAI/litellm/blob/main/LICENSE · as of 2026-10-25

Python SDK for building MCP clients and servers.

Limit not stated

License: MIT

Jobs: agent-tools

Re-fetched twice 2026-10-25 (~10 s apart, plain curl, no login): GitHub blob page HTTP 200, no redirect (page chrome differs between fetches); raw LICENSE fetched twice too, HTTP 200, byte-identical: 'MIT License' / 'Copyright (c) 2024 Anthropic, PBC'. License unchanged.

https://github.com/modelcontextprotocol/python-sdk/blob/main/LICENSE · as of 2026-10-25

Reference MCP server implementations you can run yourself.

Limit not stated

License: MIT to Apache-2.0 transition

Provider: mcp-registry

Jobs: mcp-discovery, agent-tools

Fetched LICENSE says the project is moving from MIT to Apache-2.0. Not a hosted free quota. Re-fetched twice 2026-10-25 (~10 s apart, plain curl, no login): GitHub blob page HTTP 200, no redirect (page chrome differs between fetches); raw LICENSE fetched twice too, HTTP 200, byte-identical: 'The MCP project is undergoing a licensing transition from the MIT License to the Apache License, Version 2.0'. Still states 'All new code and specification contributions to the project are licensed under Apache-2.0.' License unchanged.

https://github.com/modelcontextprotocol/servers/blob/main/LICENSE · as of 2026-10-25

Public metadata registry of MCP servers. The MCP Registry provides an unauthenticated read-only REST API (GET /v0.1/servers) at https://registry.modelcontextprotocol.io. No price is published on the fetched docs.

Limit not stated

License: MIT to Apache-2.0 transition

Provider: mcp-registry

Jobs: mcp-discovery

Docs fetched 2026-10-24: 'The MCP Registry provides an unauthenticated read-only REST API that aggregators can use to populate their data stores.' 'The MCP Registry is currently in preview.' It 'does not provide uptime or data durability guarantees'. Publishing requires namespace-based authentication (https://github.com/modelcontextprotocol/registry/blob/main/docs/reference/api/official-registry-api.md). Neither page publishes a price or a numeric rate limit, so the hosted API is not stored as a $0 price or a quota. cost oss is the registry repo LICENSE (https://github.com/modelcontextprotocol/registry/blob/main/LICENSE, fetched 2026-10-24): 'The MCP project is undergoing a licensing transition from the MIT License to the Apache License, Version 2.0'. Not a substitute for Depot picks.

https://modelcontextprotocol.io/registry/registry-aggregators · as of 2026-10-24

Open-source sandbox SDK. Hosted E2B sandbox time is not included and is not given a free quota here.

Limit not stated

License: Apache-2.0

Provider: e2b

Jobs: agent-sandbox, code-interpreter

Do not read this as a free hosted sandbox tier. Hosted rates stay in pricing rows. Re-fetched twice 2026-10-25 (~10 s apart, plain curl, no login): GitHub blob page HTTP 200, no redirect (page chrome differs between fetches); raw LICENSE fetched twice too, HTTP 200, byte-identical: 'Apache License Version 2.0, January 2004' / 'Copyright 2023 FoundryLabs, Inc.'. License unchanged.

https://github.com/e2b-dev/e2b/blob/main/LICENSE · as of 2026-10-25

Self-hosted or embedded vector database.

Limit not stated

License: Apache-2.0

Provider: chroma

Jobs: self-host-vector, prototype-rag, embedded-vector

Re-fetched twice 2026-10-25 (~10 s apart, plain curl, no login): GitHub blob page HTTP 200, no redirect (page chrome differs between fetches); raw LICENSE fetched twice too, HTTP 200, byte-identical: 'Apache License Version 2.0, January 2004'. License unchanged.

https://github.com/chroma-core/chroma/blob/main/LICENSE · as of 2026-10-25

Self-hosted vector database. Qdrant Cloud is not this row.

Limit not stated

License: Apache-2.0

Provider: qdrant

Jobs: self-host-vector, managed-vector, hybrid-search-rag

LICENSE refetched 2026-10-03 is Apache License 2.0. README fetched the same day says hybrid search combines semantic understanding and keyword precision, and that sparse vectors are for full-text search. It does not say BM25. It does not say enterprise RAG with rerank and support, agent memory across sessions, or a knowledge graph. Qdrant Cloud is not this row. Re-fetched twice 2026-10-25 (~10 s apart, plain curl, no login): GitHub blob page HTTP 200, no redirect (page chrome differs between fetches); raw LICENSE fetched twice too, HTTP 200, byte-identical: 'Apache License Version 2.0, January 2004' / 'Copyright 2026 Qdrant Solutions GmbH'. License unchanged.

https://github.com/qdrant/qdrant/blob/master/LICENSE · as of 2026-10-25

Embedded vector database library.

Limit not stated

License: Apache-2.0

Provider: lancedb

Jobs: embedded-vector, self-host-vector

Re-fetched twice 2026-10-25 (~10 s apart, plain curl, no login): GitHub blob page HTTP 200, no redirect (page chrome differs between fetches); raw LICENSE fetched twice too, HTTP 200, byte-identical: 'Apache License Version 2.0, January 2004'. License unchanged.

https://github.com/lancedb/lancedb/blob/main/LICENSE · as of 2026-10-25

Postgres extension for vector similarity. Not Supabase hosted Postgres.

Limit not stated

License: PostgreSQL License

Jobs: postgres-vector, self-host-vector

providerId is null on purpose. supabase-pgvector is a hosted product and is not this extension. Re-fetched twice 2026-10-25 (~10 s apart, plain curl, no login): GitHub blob page HTTP 200, no redirect (page chrome differs between fetches); raw LICENSE fetched twice too, HTTP 200, byte-identical: 'Portions Copyright (c) 1996-2026, PostgreSQL Global Development Group' / 'without fee, and without a written agreement is hereby granted'. License unchanged.

https://github.com/pgvector/pgvector/blob/master/LICENSE · as of 2026-10-25

In-process analytical SQL database.

Limit not stated

License: MIT

Jobs: sql-warehouse-agent-context, agent-analytics-warehouse

Copyright line fetched: 2018-2026 Stichting DuckDB Foundation. Re-fetched twice 2026-10-25 (~10 s apart, plain curl, no login): GitHub blob page HTTP 200, no redirect (page chrome differs between fetches); raw LICENSE fetched twice too, HTTP 200, byte-identical: 'Copyright 2018-2026 Stichting DuckDB Foundation' / 'Permission is hereby granted, free of charge'. License unchanged.

https://github.com/duckdb/duckdb/blob/main/LICENSE · as of 2026-10-25

Embedded SQL database engine and its documentation, dedicated to the public domain.

Limit not stated

License: Public Domain

Jobs: edge-sqlite-agent-db

Re-fetched twice 2026-10-25 (~10 s apart, plain curl, no login): HTTP 200, no redirect, byte-identical: 'SQLite Is Public Domain' / 'All of the code and documentation in SQLite has been dedicated to the public domain by the authors.'. License unchanged.

https://www.sqlite.org/copyright.html · as of 2026-10-25

Library that converts common office and web files to Markdown for agents.

Limit not stated

License: MIT

Jobs: document-parse, llm-ready-extract

Re-fetched twice 2026-10-25 (~10 s apart, plain curl, no login): GitHub blob page HTTP 200, no redirect (page chrome differs between fetches); raw LICENSE fetched twice too, HTTP 200, byte-identical: 'MIT License' / 'Copyright (c) Microsoft Corporation.'. License unchanged.

https://github.com/microsoft/markitdown/blob/main/LICENSE · as of 2026-10-25

Model loading and training library. Weights keep their own licenses.

Limit not stated

License: Apache-2.0

Jobs: open-model-inference, finetune

Re-fetched twice 2026-10-25 (~10 s apart, plain curl, no login): GitHub blob page HTTP 200, no redirect (page chrome differs between fetches); raw LICENSE fetched twice too, HTTP 200, byte-identical: 'Copyright 2018- The Hugging Face team. All rights reserved.' / 'Apache License Version 2.0, January 2004'. License unchanged.

https://github.com/huggingface/transformers/blob/main/LICENSE · as of 2026-10-25

Posts lint findings as review comments on a code host when the findings are in the diff under review. It can also filter lint output by diff locally. It does not author the review.

Limit not stated

License: MIT

Jobs: diff-aware-pr-review

LICENSE fetched 2026-10-03 is MIT, copyright 2016 haya14busa. README fetched the same day says it posts a comment if findings are in the diff of patches to review. Re-fetched twice 2026-10-25 (~10 s apart, plain curl, no login): HTTP 200, no redirect, byte-identical: 'MIT License' / 'Copyright (c) 2016 haya14busa'. License unchanged.

https://raw.githubusercontent.com/reviewdog/reviewdog/master/LICENSE · as of 2026-10-25

Engine that runs a delivery task in containers on a laptop, a CI server, or the cloud. The README says tools run in containers orchestrated by sandboxed functions, and that the same task can run in an AI sandbox. You still supply a container runtime. Not a hosted GPU runner.

Limit not stated

License: Apache-2.0

Jobs: agent-ephemeral-ci

LICENSE fetched 2026-10-03 is Apache License 2.0. README fetched the same day. No price is stored. Re-fetched twice 2026-10-25 (~10 s apart, plain curl, no login): HTTP 200, no redirect, byte-identical: 'Apache License Version 2.0, January 2004' / 'Copyright 2022 Dagger, Inc.'. License unchanged.

https://raw.githubusercontent.com/dagger/dagger/main/LICENSE · as of 2026-10-25

The specification for using a container as a coding environment. devcontainer.json stores the tools, libraries, and setup for working inside a local or cloud container. This row is the spec, not a running workspace. The fetched README does not say files remain after the container stops.

Limit not stated

License: CC-BY-4.0

Jobs: persistent-dev-env

LICENSE fetched 2026-10-03 begins Attribution 4.0 International. README says copyright Microsoft Corporation and Creative Commons Attribution 4.0. Re-fetched twice 2026-10-25 (~10 s apart, plain curl, no login): HTTP 200, no redirect, byte-identical: 'Attribution 4.0 International' / 'Creative Commons Attribution 4.0 International Public License'. License unchanged.

https://raw.githubusercontent.com/devcontainers/spec/main/LICENSE · as of 2026-10-25

Library for long-running stateful agents. The README says agents can persist through failures, resume, and run for extended periods, and that a person can inspect and modify agent state. This is not a hosted LangGraph runtime.

Limit not stated

License: MIT

Jobs: durable-agent-workflows

LICENSE fetched 2026-10-03 is MIT, copyright 2024 LangChain, Inc. README fetched the same day. Its production-deployment link points at LangSmith Deployment, which is not this row. Re-fetched twice 2026-10-25 (~10 s apart, plain curl, no login): HTTP 200, no redirect, byte-identical: 'MIT License' / 'Copyright (c) 2024 LangChain, Inc.'. License unchanged.

https://raw.githubusercontent.com/langchain-ai/langgraph/main/LICENSE · as of 2026-10-25

Self-hostable engine for background tasks and AI agents. The README says event-based triggering is for event-driven systems, and that tasks have configurable retries. It does not say debounce or fan-out. Hatchet Cloud is not this row.

Limit not stated

License: MIT

Provider: hatchet

Jobs: event-driven-agent-jobs

LICENSE fetched 2026-10-03 is MIT, copyright 2023-present Hatchet Technologies Inc. README fetched the same day. Re-fetched twice 2026-10-25 (~10 s apart, plain curl, no login): HTTP 200, no redirect, byte-identical: 'MIT License' / 'Copyright (c) 2023-present Hatchet Technologies Inc.'. License unchanged.

https://raw.githubusercontent.com/hatchet-dev/hatchet/main/LICENSE · as of 2026-10-25

Distributed runtime for scaling Python and AI applications from a laptop to a cluster. The README does not describe a managed control plane. Anyscale is not this row.

Limit not stated

License: Apache-2.0

Jobs: ray-distributed-agents

LICENSE fetched 2026-10-03 is Apache License 2.0. README.rst fetched the same day says Ray is a unified framework for scaling AI and Python applications. Re-fetched twice 2026-10-25 (~10 s apart, plain curl, no login): HTTP 200, no redirect, byte-identical: 'Apache License Version 2.0, January 2004' / 'Copyright 2023 Ray Authors'. The file continues with notices for bundled third-party code (no extra terms on Ray itself). License unchanged.

https://raw.githubusercontent.com/ray-project/ray/master/LICENSE · as of 2026-10-25

Eval toolkit. The README says the rag_eval template evaluates RAG systems and that Ragas provides objective metrics for LLM apps. The fetched README does not name faithfulness, context precision, or context recall.

Limit not stated

License: Apache-2.0

Provider: ragas

Jobs: rag-evals

LICENSE fetched 2026-10-03 from vibrantlabsai/ragas is Apache License 2.0. README fetched the same day. Re-fetched twice 2026-10-25 (~10 s apart, plain curl, no login): HTTP 200, no redirect, byte-identical: 'Apache License Version 2.0, January 2004' / 'Copyright [2023] [Vibrant Labs]'. License unchanged.

https://raw.githubusercontent.com/vibrantlabsai/ragas/main/LICENSE · as of 2026-10-25

Python framework. The README says input and output guards detect, quantify, and mitigate specific risks. The fetched README does not say hallucination monitoring or research-grade metrics. Hosted remote inferencing is not this row.

Limit not stated

License: Apache-2.0

Jobs: production-guardrail-evals

LICENSE fetched 2026-10-03 is Apache License 2.0. README fetched the same day says hosted remote inferencing is being discontinued, planned cutoff August 25, 2026. Re-fetched twice 2026-10-25 (~10 s apart, plain curl, no login): HTTP 200, no redirect, byte-identical: 'Apache License Version 2.0, January 2004'. README.md (https://raw.githubusercontent.com/guardrails-ai/guardrails/main/README.md) re-fetched twice 2026-10-25 (~10 s apart, plain curl, no login): HTTP 200, no redirect, byte-identical; it still says Guardrails is discontinuing its hosted remote inferencing, 'Planned cutoff: August 25, 2026.' (cutoff unchanged). License unchanged.

https://raw.githubusercontent.com/guardrails-ai/guardrails/main/LICENSE · as of 2026-10-25

TypeScript task platform. The README lists durable tasks, retries, queues, and idempotency for agents and AI applications. It does not describe idempotency keys or exactly-once charges, tickets, or emails. No price is stored.

Limit not stated

License: Apache-2.0

Provider: trigger-dev

Jobs: durable-idempotency

LICENSE fetched 2026-10-03 is Apache License 2.0. README fetched the same day. Re-fetched twice 2026-10-25 (~10 s apart, plain curl, no login): HTTP 200, no redirect, byte-identical: 'Apache License Version 2.0, January 2004' / 'Copyright [2023] [Trigger.dev]'. License unchanged.

https://raw.githubusercontent.com/triggerdotdev/trigger.dev/main/LICENSE · as of 2026-10-25

Self-hosted observability. The README says the community edition runs in your infrastructure and includes logs, metrics, traces, alerts, exceptions, and span views. Directories ee/ and cmd/enterprise/ use another license and are not this row. SigNoz Cloud is not this row.

Limit not stated

License: MIT Expat outside ee/

Jobs: agent-error-observability

LICENSE fetched 2026-10-03 says content outside ee/ and cmd/enterprise/ is MIT Expat. README fetched the same day. Re-fetched twice 2026-10-25 (~10 s apart, plain curl, no login): HTTP 200, no redirect, byte-identical: 'Copyright (c) 2020-present SigNoz Inc.' / '"MIT Expat" license'. Directories named in the LICENSE are still 'ee/' and 'cmd/enterprise/'. License unchanged.

https://raw.githubusercontent.com/SigNoz/signoz/main/LICENSE · as of 2026-10-25

Open-source memory library. The README says it retains user, session, and agent state, and that a browser extension stores memories across ChatGPT, Perplexity, and Claude. It does not say MCP. Benchmark figures in the README are for the managed platform, not this row.

Limit not stated

License: Apache-2.0

Provider: mem0

Jobs: cross-session-user-memory

LICENSE fetched 2026-10-03 is Apache License 2.0. README fetched the same day. Re-fetched twice 2026-10-25 (~10 s apart, plain curl, no login): HTTP 200, no redirect, byte-identical: 'Apache License Version 2.0, January 2004' / 'Copyright [2023] [Taranjeet Singh]'. License unchanged.

https://raw.githubusercontent.com/mem0ai/mem0/main/LICENSE · as of 2026-10-25

Framework for temporal context graphs for AI agents. The README says facts have validity windows for what is true now and what was true before, and that episodes are the raw data those facts come from. Managed Zep is not this row.

Limit not stated

License: Apache-2.0

Jobs: episodic-long-term-memory

LICENSE fetched 2026-10-03 is Apache License 2.0. README fetched the same day. Re-fetched twice 2026-10-25 (~10 s apart, plain curl, no login): HTTP 200, no redirect, byte-identical: 'Apache License Version 2.0, January 2004'. License unchanged.

https://raw.githubusercontent.com/getzep/graphiti/main/LICENSE · as of 2026-10-25

HTTP pub-sub server. The README says scripts can send notifications to a phone or desktop, and that open-source Android and iOS apps exist. It does not mention agents. ntfy Pro is paid and is not this row.

Limit not stated

License: Apache-2.0

Jobs: agent-push-notifications

LICENSE fetched 2026-10-03 is Apache License 2.0. README fetched the same day also says ntfy.sh can be used without signing up. That hosted site is not priced here. Re-fetched twice 2026-10-25 (~10 s apart, plain curl, no login): HTTP 200, no redirect, byte-identical: 'Apache License Version 2.0, January 2004' / 'Copyright 2021 Philipp C. Heckel'. License unchanged.

https://raw.githubusercontent.com/binwiederhier/ntfy/main/LICENSE · as of 2026-10-25

Real-time messaging server. The README says clients connect over WebSocket or Server-Sent Events, subscribe to channels, and that presence and AI streaming responses are uses. It does not mention agent tool progress.

Limit not stated

License: Apache-2.0

Jobs: realtime-agent-channel

LICENSE fetched 2026-10-03 is Apache License 2.0. README fetched the same day. Re-fetched twice 2026-10-25 (~10 s apart, plain curl, no login): HTTP 200, no redirect, byte-identical: 'Apache License Version 2.0, January 2004' / 'Copyright 2023 Centrifugal Labs LTD'. License unchanged.

https://raw.githubusercontent.com/centrifugal/centrifugo/master/LICENSE · as of 2026-10-25

SDK whose README says agents get pre-authenticated toolkits and a hosted MCP URL from a session. A COMPOSIO_API_KEY is required. The fetched README does not name Gmail, GitHub, or a CRM. Hosted toolkit pricing is not stored.

Limit not stated

License: MIT

Provider: composio

Jobs: agent-tool-router

LICENSE fetched 2026-10-03 is MIT, copyright 2025 Sampark Inc. README fetched the same day. Re-fetched twice 2026-10-25 (~10 s apart, plain curl, no login): HTTP 200, no redirect, byte-identical: 'MIT License' / 'Copyright (c) 2025 Sampark Inc.'. License unchanged.

https://raw.githubusercontent.com/ComposioHQ/composio/master/LICENSE · as of 2026-10-25

Local Python input scanners, including one named PromptInjection. The README says the project is archived. This row is not a hosted firewall.

Limit not stated

License: MIT

Jobs: prompt-injection-firewall

LICENSE fetched 2026-10-03 is the MIT License, copyright Protect AI. README fetched the same day lists PromptInjection under prompt scanners. It does not say firewall, and it does not say the scanner checks retrieved documents. Re-fetched twice 2026-10-25 (~10 s apart, plain curl, no login): HTTP 200, no redirect, byte-identical: 'The MIT License (MIT)' / 'Copyright (c) Protect AI. All rights reserved.'. License unchanged.

https://raw.githubusercontent.com/protectai/llm-guard/main/LICENSE · as of 2026-10-25

Public software-engineering benchmark instances. Download from the GitHub repo or the Hugging Face dataset princeton-nlp/SWE-bench.

Limit not stated

License: MIT

Jobs: oss-evals, coding-agent

Dataset card: https://huggingface.co/datasets/princeton-nlp/SWE-bench. Instance text can include upstream repository code under other licenses. Re-fetched twice 2026-10-25: https://github.com/princeton-nlp/SWE-bench/blob/main/LICENSE now answers HTTP 301 to https://github.com/SWE-bench/SWE-bench/blob/main/LICENSE (HTTP 200; the GitHub blob page changes between fetches, page chrome), so sourceUrl moved to the redirect target; the raw file https://raw.githubusercontent.com/SWE-bench/SWE-bench/main/LICENSE was also fetched twice (HTTP 200, byte-identical): 'MIT License' / 'Copyright (c) 2023 Carlos E Jimenez, John Yang, Alexander Wettig, Shunyu Yao, Kexin Pei, Ofir Press, Karthik R Narasimhan'.

https://github.com/SWE-bench/SWE-bench/blob/main/LICENSE · as of 2026-10-25

Gated public benchmark of tool-using assistant questions. Files require accepting a no-reshare condition. No license field was on the fetched card.

Limit not stated

Jobs: evals, oss-evals, deep-research-agents

README fetched 2026-10-03. extra_gated_prompt says not to reshare outside a gated or private HF repo. No license key in the fetched card. Anonymous file download returned 401 earlier on the raw README path before this resolve URL. README re-fetched twice 2026-10-25 (HTTP 200, byte-identical): front matter still has extra_gated_prompt 'To avoid contamination and data leakage, you agree to not reshare this dataset outside of a gated or private repository on the HF hub.' and still no license key, so license stays null.

https://huggingface.co/datasets/gaia-benchmark/GAIA/resolve/main/README.md · as of 2026-10-25

Public cais/mmlu multiple-choice eval dataset.

Limit not stated

License: MIT

Jobs: evals, oss-evals

Dataset card fetched 2026-10-03 says license: mit. The hendrycks/test GitHub LICENSE fetched the same day is also MIT. Dataset card re-fetched twice 2026-10-25 (HTTP 200, byte-identical): front matter 'license:' / '- mit'.

https://huggingface.co/datasets/cais/mmlu/raw/main/README.md · as of 2026-10-25

Public MMLU-Pro eval dataset, a harder multiple-choice successor set.

Limit not stated

License: MIT

Jobs: evals, oss-evals

Dataset card fetched 2026-10-03 says license: mit. The TIGER-AI-Lab/MMLU-Pro GitHub LICENSE fetched the same day is Apache-2.0 for the code repo; this row follows the dataset card, not the code license. Dataset card re-fetched twice 2026-10-25 (HTTP 200, byte-identical): front matter 'license: mit'.

https://huggingface.co/datasets/TIGER-Lab/MMLU-Pro/raw/main/README.md · as of 2026-10-25

Public function-calling eval dataset used by the Berkeley leaderboard.

Limit not stated

License: Apache-2.0

Jobs: evals, oss-evals, agent-tools

Dataset card fetched 2026-10-03 says license: apache-2.0. The gorilla GitHub LICENSE fetched the same day is also Apache-2.0. Dataset card re-fetched twice 2026-10-25 (HTTP 200, byte-identical): front matter 'license: apache-2.0'.

https://huggingface.co/datasets/gorilla-llm/Berkeley-Function-Calling-Leaderboard/raw/main/README.md · as of 2026-10-25

Public tau-bench agent tool-use benchmark repository.

Limit not stated

License: MIT

Jobs: evals, oss-evals, agent-tools

LICENSE fetched 2026-10-03 is MIT, copyright 2024 Sierra. LICENSE re-fetched twice 2026-10-25 (HTTP 200, byte-identical): 'MIT License' / 'Copyright (c) 2024 Sierra'.

https://raw.githubusercontent.com/sierra-research/tau-bench/main/LICENSE · as of 2026-10-25

Public terminal-bench tasks for agents that use a shell.

Limit not stated

License: Apache-2.0

Jobs: evals, oss-evals, coding-agent

LICENSE fetched 2026-10-03 is Apache License 2.0. LICENSE re-fetched twice 2026-10-25 (HTTP 200, byte-identical): 'Apache License Version 2.0, January 2004'.

https://raw.githubusercontent.com/laude-institute/terminal-bench/main/LICENSE · as of 2026-10-25

Public WebArena web-agent benchmark code and task definitions.

Limit not stated

License: Apache-2.0

Jobs: evals, web-agent, browser-agent-loop

LICENSE fetched 2026-10-03 is Apache License 2.0. Self-hosted websites in the benchmark are a separate setup, not priced here. LICENSE re-fetched twice 2026-10-25 (HTTP 200, byte-identical): 'Apache License Version 2.0, January 2004'.

https://raw.githubusercontent.com/web-arena-x/webarena/main/LICENSE · as of 2026-10-25

Public VisualWebArena benchmark, a multimodal successor to WebArena.

Limit not stated

License: MIT

Jobs: evals, web-agent, multimodal

LICENSE fetched 2026-10-03 begins with a free-of-charge permission grant (MIT). Copyright 2024 Jing Yu Koh, Robert Lo, Lawrence Jang, Vikram Duvvur, Ming Chong Lim, and Po-Yu Huang. LICENSE re-fetched twice 2026-10-25 (HTTP 200, byte-identical): 'Copyright (c) 2024 Jing Yu Koh, Robert Lo, Lawrence Jang, Vikram Duvvur, Ming Chong Lim, and Po-Yu Huang' / 'Permission is hereby granted, free of charge'.

https://raw.githubusercontent.com/web-arena-x/visualwebarena/main/LICENSE · as of 2026-10-25

Public OSWorld desktop computer-use benchmark.

Limit not stated

License: Apache-2.0

Jobs: evals, desktop-computer-use

LICENSE fetched 2026-10-03 is Apache License 2.0. LICENSE re-fetched twice 2026-10-25 (HTTP 200, byte-identical): 'Apache License Version 2.0, January 2004'.

https://raw.githubusercontent.com/xlang-ai/OSWorld/main/LICENSE · as of 2026-10-25

Public Mind2Web web-agent demonstration dataset.

Limit not stated

License: CC-BY-4.0

Jobs: evals, web-agent, form-filling-rpa

Dataset card fetched 2026-10-03 says license: cc-by-4.0. A GitHub LICENSE path returned 404. Dataset card fetched 2026-10-19 says Mind2Web is a dataset for developing and evaluating generalist agents for the web. Dataset card re-fetched twice 2026-10-25 (HTTP 200, byte-identical): front matter 'license: cc-by-4.0'.

https://huggingface.co/datasets/osunlp/Mind2Web/raw/main/README.md · as of 2026-10-25

Public AgentBench multi-environment agent evaluation repository.

Limit not stated

License: Apache-2.0

Jobs: evals, oss-evals

LICENSE fetched 2026-10-03 is Apache License 2.0. LICENSE re-fetched twice 2026-10-25 (HTTP 200, byte-identical), no redirect from the THUDM path: 'Apache License Version 2.0, January 2004'.

https://raw.githubusercontent.com/THUDM/AgentBench/main/LICENSE · as of 2026-10-25

Public SWE-bench Verified subset. The dataset card fetched here does not state a license.

Limit not stated

Jobs: evals, oss-evals, coding-agent

Hugging Face dataset API fetched 2026-10-03. cardData.license is null and gated is false. The existing swe-bench row is the MIT code license, not this subset card. License left null. Dataset API re-fetched twice 2026-10-25 (HTTP 200, byte-identical): cardData.license still null and gated false, so license stays null.

https://huggingface.co/api/datasets/SWE-bench/SWE-bench_Verified · as of 2026-10-25

Public BrowserGym environment for web agents.

Limit not stated

License: Apache-2.0

Jobs: evals, web-agent, self-host-browser

LICENSE fetched 2026-10-03 is Apache License 2.0, copyright 2024 ServiceNow. README fetched 2026-10-19 says it is a framework to implement, test, and evaluate web agents on BrowserGym benchmarks. LICENSE re-fetched twice 2026-10-25 (HTTP 200, byte-identical): 'Copyright 2024 ServiceNow' / 'Licensed under the Apache License, Version 2.0'.

https://raw.githubusercontent.com/ServiceNow/BrowserGym/main/LICENSE · as of 2026-10-25

Public OpenAI HumanEval code-generation dataset.

Limit not stated

License: MIT

Jobs: evals, coding-agent

Dataset card fetched 2026-10-03 says license: mit. openai/human-eval GitHub LICENSE fetched the same day is also MIT. Dataset card re-fetched twice 2026-10-25 (HTTP 200, byte-identical): front matter 'license:' / '- mit'.

https://huggingface.co/datasets/openai/openai_humaneval/raw/main/README.md · as of 2026-10-25

Public GSM8K grade-school math word problems.

Limit not stated

License: MIT

Jobs: evals

Dataset card fetched 2026-10-03 says license: mit. openai/grade-school-math GitHub LICENSE fetched the same day is also MIT. Dataset card fetched 2026-10-19 says it has been used for many benchmarks, including the Open LLM Leaderboard. Dataset card re-fetched twice 2026-10-25 (HTTP 200, byte-identical): front matter 'license:' / '- mit'.

https://huggingface.co/datasets/openai/gsm8k/raw/main/README.md · as of 2026-10-25

Public WildChat conversation traces. The card license is Open Data Commons Attribution (ODC-By).

Limit not stated

License: ODC-By

Jobs: finetune

Dataset card fetched 2026-10-03 says license: odc-by. These are chat traces, not a scored leaderboard. Dataset card fetched 2026-10-19 lists the task category instruction-finetuning and describes user-chatbot conversations not covered by other instruction fine-tuning datasets; it does not describe an eval set, so it is tagged finetune, not evals. Re-fetched twice 2026-10-25: https://huggingface.co/datasets/allenai/wildchat/raw/main/README.md now answers HTTP 307 to /datasets/allenai/WildChat/raw/main/README.md (HTTP 200, byte-identical), so sourceUrl moved to the redirect target; front matter still 'license: odc-by'.

https://huggingface.co/datasets/allenai/WildChat/raw/main/README.md · as of 2026-10-25

Public Toucan 1.5M tool-use trajectory dataset.

Limit not stated

License: Apache-2.0

Jobs: agent-tools, synthetic-eval-data

README fetched 2026-10-03 says license: apache-2.0 and 'This dataset is released under Apache 2.0.' README fetched 2026-10-19 describes it as a fully synthetic tool-agent dataset of 1.5M trajectories used for supervised fine-tuning; it is training data, not an eval set, so it is not tagged evals. README re-fetched twice 2026-10-25 (HTTP 200, byte-identical): front matter 'license: apache-2.0' and 'This dataset is released under Apache 2.0.'

https://huggingface.co/datasets/Agent-Ark/Toucan-1.5M/raw/main/README.md · as of 2026-10-25

Public OpenAI MRCR long-context multi-needle benchmark dataset.

Limit not stated

License: MIT

Jobs: evals, long-context-agents

Dataset card fetched 2026-10-03 says license: mit. Dataset card re-fetched twice 2026-10-25 (HTTP 200, byte-identical): front matter 'license: mit'.

https://huggingface.co/datasets/openai/mrcr/raw/main/README.md · as of 2026-10-25

Public TheAgentCompany simulated-workplace agent benchmark.

Limit not stated

License: MIT

Jobs: evals, oss-evals

LICENSE fetched 2026-10-03 is MIT, copyright 2024 TheAgentCompany. LICENSE re-fetched twice 2026-10-25 (HTTP 200, byte-identical): 'MIT License' / 'Copyright (c) 2024 TheAgentCompany'.

https://raw.githubusercontent.com/TheAgentCompany/TheAgentCompany/main/LICENSE · as of 2026-10-25

Public multi-language SWE-bench style issue-resolution benchmark.

Limit not stated

License: Apache-2.0

Jobs: evals, coding-agent

LICENSE fetched 2026-10-03 is Apache License 2.0. LICENSE re-fetched twice 2026-10-25 (HTTP 200, byte-identical): 'Apache License Version 2.0, January 2004'.

https://raw.githubusercontent.com/multi-swe-bench/multi-swe-bench/main/LICENSE · as of 2026-10-25

The MCP specification, including tool and schema shapes, published at no charge.

Limit not stated

License: Apache-2.0

Provider: mcp-registry

Jobs: agent-tools, mcp-discovery

Spec revision date is 2026-07-28. Repo LICENSE text fetched for the servers and registry repos describes an MIT to Apache-2.0 transition; this row does not restate unpublished draft text. Re-fetched twice 2026-10-25 (~10 s apart, plain curl, no login): the spec page answered HTTP 200, byte-identical (title 'Specification - Model Context Protocol', navigation links /specification/2026-07-28/basic/index; the page itself carries no license text). The spec repo LICENSE https://raw.githubusercontent.com/modelcontextprotocol/modelcontextprotocol/main/LICENSE was also fetched twice (HTTP 200, byte-identical): 'All new code and specification contributions to the project are licensed under Apache-2.0.' License unchanged.

https://modelcontextprotocol.io/specification/2026-07-28 · as of 2026-10-25

The public llms.txt proposal for agent-readable site maps.

Limit not stated

No license and no quota are stated on the fetched page. This is the convention, not Depot's own /llms.txt. Re-fetched twice 2026-10-25 (~10 s apart, plain curl, no login): HTTP 200, no redirect, byte-identical (title 'The /llms.txt file, v2 – llms-txt'); still no license and no quota stated on the page, so license and limits stay null.

https://llmstxt.org/ · as of 2026-10-25

The OpenAPI specification for describing HTTP APIs.

Limit not stated

License: Apache-2.0

Jobs: agent-tools

Published spec: https://spec.openapis.org/oas/latest.html Re-fetched twice 2026-10-25 (~10 s apart, plain curl, no login): the GitHub blob page answered HTTP 200 both times (page chrome differs between fetches), so the raw file https://raw.githubusercontent.com/OAI/OpenAPI-Specification/main/LICENSE was also fetched twice (HTTP 200, byte-identical): 'Apache License Version 2.0, January 2004'. License unchanged.

https://github.com/OAI/OpenAPI-Specification/blob/main/LICENSE · as of 2026-10-25

Public API reference and guides. Reading them is free. API calls are not.

Limit not stated

Provider: openai

Jobs: agent-tools

No free inference quota is stated by this row. Signup credits were not fetched and are not listed. Re-fetched twice 2026-10-25 (~10 s apart, plain curl, no login): https://platform.openai.com/docs now answers HTTP 301 to https://developers.openai.com/api/docs/overview, then HTTP 301 to /api/docs (HTTP 200, byte-identical, title 'OpenAI API Platform Documentation'), so sourceUrl moved to https://developers.openai.com/api/docs. The page still states no free inference quota; license and limits stay null.

https://developers.openai.com/api/docs · as of 2026-10-25

Public API reference and guides. Reading them is free. API calls are not.

Limit not stated

Provider: anthropic

Jobs: agent-tools

No free inference quota is stated by this row. Re-fetched twice 2026-10-25 (~10 s apart, plain curl, no login): https://docs.anthropic.com now answers HTTP 301 to https://platform.claude.com/docs/, then HTTP 308 to /docs and HTTP 307 to /docs/en/home (HTTP 200, title 'Documentation - Claude Platform Docs'; the page payload differs between fetches, the title does not), so sourceUrl moved to https://platform.claude.com/docs/en/home. The page still states no free inference quota; license and limits stay null.

https://platform.claude.com/docs/en/home · as of 2026-10-25

Public Groq API docs, including the rate-limits page used for the Free Plan rows.

Limit not stated

Provider: groq

Jobs: ultra-low-latency

Page fetched twice 2026-10-25 about 10 s apart without logging in (HTTP 200, no redirect, response header 'cache-control: public, max-age=0, must-revalidate', identical HTML; the page shows no last-updated date). It is the public Rate Limits page of the Groq docs (page title 'Rate Limits - GroqDocs'; same docs navigation lists Quickstart, Models and API Reference). 'Rate limits act as control measures to regulate how frequently users and applications can access our API within specified timeframes.' 'Rate limits apply at the organization level, not individual users.' 'The following is a high level summary and there may be exceptions to these limits.' 'You can view the current, exact rate limits for your organization on the limits page in your account settings.' The page has a 'Free Plan Limits' tab next to 'Developer Plan Limits' (the groq-free-plan rows are quoted from it). Docs access is not the Free Plan quota. Use the groq-free-plan rows for numbers; limits stays null on this row.

https://console.groq.com/docs/rate-limits · as of 2026-10-25

Agent2Agent protocol specification and repository.

Limit not stated

License: Apache-2.0

Jobs: a2a-agent-discovery

LICENSE fetched 2026-10-03 is Apache License 2.0. Spec site https://a2a-protocol.org/latest/ also returned 200 the same day. Re-fetched twice 2026-10-25 (~10 s apart, plain curl, no login): HTTP 200, byte-identical: 'Apache License Version 2.0, January 2004'. License unchanged. Spec site https://a2a-protocol.org/latest/ re-fetched twice 2026-10-25 (~10 s apart, plain curl, no login): HTTP 200, no redirect, byte-identical; page title 'A2A Protocol', text 'Agent2Agent (A2A) protocol'.

https://raw.githubusercontent.com/a2aproject/A2A/main/LICENSE · as of 2026-10-25

AGENTS.md convention for telling coding agents how a repo works.

Limit not stated

License: MIT

Jobs: coding-agent

LICENSE fetched 2026-10-03 is MIT, copyright 2025 OpenAI. https://agents.md/ returned 200 the same day. Re-fetched twice 2026-10-25 (~10 s apart, plain curl, no login): HTTP 200, byte-identical: 'MIT License' / 'Copyright (c) 2025 OpenAI'. License unchanged. https://agents.md/ re-fetched twice 2026-10-25 (~10 s apart, plain curl, no login): HTTP 200, no redirect, byte-identical; page title 'AGENTS.md', text 'AGENTS.md is a simple, open format for guiding coding agents'.

https://raw.githubusercontent.com/agentsmd/agents.md/main/LICENSE · as of 2026-10-25

Public Agent Skills docs index, including the specification links.

Limit not stated

Jobs: coding-agent, mcp-discovery

llms.txt fetched 2026-10-03 lists the Agent Skills docs. No license line was in that file. The HTML specification page also returned 200. License left null. Re-fetched twice 2026-10-25 (~10 s apart, plain curl, no login): HTTP 200, byte-identical; the file still starts '# Agent Skills' and links [Specification](https://agentskills.io/specification.md), and still has no license line, so license stays null.

https://agentskills.io/llms.txt · as of 2026-10-25

Open-source OCR toolkit you run yourself.

Limit not stated

License: Apache-2.0

Jobs: layout-aware-ocr, document-parse

LICENSE fetched 2026-10-03 is Apache License 2.0. Re-fetched twice 2026-10-25 (~10 s apart, plain curl, no login): HTTP 200, byte-identical: 'Apache License Version 2.0, January 2004'. License unchanged.

https://raw.githubusercontent.com/PaddlePaddle/PaddleOCR/main/LICENSE · as of 2026-10-25

Open-source OCR library you run yourself.

Limit not stated

License: Apache-2.0

Jobs: layout-aware-ocr, document-parse

LICENSE fetched 2026-10-03 is Apache License 2.0. Re-fetched twice 2026-10-25 (~10 s apart, plain curl, no login): HTTP 200, byte-identical: 'Apache License Version 2.0, January 2004'. License unchanged.

https://raw.githubusercontent.com/JaidedAI/EasyOCR/master/LICENSE · as of 2026-10-25

Open-source OCR engine you run yourself.

Limit not stated

License: Apache-2.0

Jobs: layout-aware-ocr, document-parse

LICENSE fetched 2026-10-03 is Apache License 2.0. Re-fetched twice 2026-10-25 (~10 s apart, plain curl, no login): HTTP 200, byte-identical: 'Apache License Version 2.0, January 2004'. License unchanged.

https://raw.githubusercontent.com/tesseract-ocr/tesseract/main/LICENSE · as of 2026-10-25

Open-source document converter to structured text for LLM input.

Limit not stated

License: MIT

Jobs: document-parse, layout-aware-ocr

LICENSE fetched 2026-10-03 is MIT, copyright The Docling Contributors. Re-fetched twice 2026-10-25 (~10 s apart, plain curl, no login): HTTP 200, byte-identical: 'MIT License' / 'Copyright The Docling Contributors'. License unchanged.

https://raw.githubusercontent.com/docling-project/docling/main/LICENSE · as of 2026-10-25

Open-source PDF-to-markdown converter you run yourself. This row is the code license, not a hosted API.

Limit not stated

License: Apache-2.0

Jobs: document-parse

LICENSE fetched 2026-10-03 is Apache License 2.0. Re-fetched twice 2026-10-25 (~10 s apart, plain curl, no login): HTTP 200, byte-identical: 'Apache License Version 2.0, January 2004'. License unchanged.

https://raw.githubusercontent.com/datalab-to/marker/master/LICENSE · as of 2026-10-25

Open-source document parser. Apache-2.0 plus extra terms: a separate commercial license is required above a stated MAU or revenue threshold.

{"commercialLicenseIfMonthlyActiveUsersExceed":100000000,"commercialLicenseIfMonthlyRevenueUsdExceeds":20000000}

License: Apache-2.0 plus additional commercial terms

Jobs: document-parse, layout-aware-ocr

LICENSE.md fetched 2026-10-03. Commercial use is allowed without a separate license until monthly active users exceed 100 million or monthly revenue exceeds USD 20 million. Online services must attribute MinerU. Not stored as plain Apache. Re-fetched twice 2026-10-25 (~10 s apart, plain curl, no login): HTTP 200, byte-identical: 'MinerU is licensed under Apache License 2.0 and is subject to the additional terms below.', 'monthly active users (MAU) exceed 100 million', 'total monthly revenue exceeds USD 20 million', and the online-service attribution term ('that MinerU is used'). License and limits unchanged.

https://raw.githubusercontent.com/opendatalab/MinerU/master/LICENSE.md · as of 2026-10-25

Open-source document parsing library you run yourself. This row is not the hosted Unstructured API.

Limit not stated

License: Apache-2.0

Jobs: document-parse

LICENSE.md fetched 2026-10-03 is Apache License 2.0. Re-fetched twice 2026-10-25 (~10 s apart, plain curl, no login): HTTP 200, byte-identical: 'Apache License Version 2.0, January 2004'. License unchanged.

https://raw.githubusercontent.com/Unstructured-IO/unstructured/main/LICENSE.md · as of 2026-10-25

Open-source crawler and browser toolkit you run yourself.

Limit not stated

License: Apache-2.0 plus attribution requirement

Jobs: self-host-browser, web-agent, deep-research-agents, web-scrape

LICENSE refetched 2026-10-03 is Apache License 2.0. README fetched the same day says Crawl4AI turns websites into clean, LLM-ready Markdown and is an open-source crawler and scraper you run yourself. The hosted cloud offer is not this row. Re-fetched twice 2026-10-25 (~10 s apart, plain curl, no login): HTTP 200, no redirect, byte-identical: 'Apache License Version 2.0, January 2004'. FLAG for review: the same LICENSE ends with an 'Attribution Requirement' section ('All distributions, publications, or public uses of this software, or derivative works based on this software, must include the following attribution'), which the stored license value 'Apache-2.0' does not capture; the GitHub commits API lists the last LICENSE change on 2025-02-21, so the term predates the 2026-10-03 fetch. License value left as stored (not re-labelled) pending that review. README.md (also fetched twice, HTTP 200, byte-identical) still says 'open-source web crawler for LLMs and AI agents' and 'LLM-ready Markdown'. License relabelled 2026-10-25 (v0.12.118): raw LICENSE re-fetched twice again (~10 s apart, plain curl, no login): HTTP 200, no redirect, byte-identical; it still opens 'Apache License Version 2.0, January 2004' and still ends with the 'Attribution Requirement' section ('All distributions, publications, or public uses of this software, or derivative works based on this software, must include the following attribution', naming UncleCode and the Crawl4AI project; for websites/web applications 'In an "About" or "Credits" section'). License value 'Apache-2.0' -> 'Apache-2.0 plus attribution requirement' (same style as mineru's 'Apache-2.0 plus additional commercial terms'); data decision by Agent Depo under Ethan delegation 2026-10-21 19:41 ET t55u. Flag resolved.

https://raw.githubusercontent.com/unclecode/crawl4ai/main/LICENSE · as of 2026-10-25

Open-source Firefox-based browser build for automation.

Limit not stated

License: MPL-2.0

Jobs: self-host-browser, computer-use-browser

LICENSE fetched 2026-10-03 is Mozilla Public License 2.0. Re-fetched twice 2026-10-25 (~10 s apart, plain curl, no login): HTTP 200, no redirect, byte-identical: 'Mozilla Public License Version 2.0'. License unchanged.

https://raw.githubusercontent.com/daijro/camoufox/main/LICENSE · as of 2026-10-25

Open-source headless browser aimed at automation.

Limit not stated

License: AGPL-3.0

Jobs: self-host-browser, web-agent

LICENSE fetched 2026-10-03 is GNU Affero GPL version 3. Re-fetched twice 2026-10-25 (~10 s apart, plain curl, no login): HTTP 200, no redirect, byte-identical: 'GNU AFFERO GENERAL PUBLIC LICENSE Version 3, 19 November 2007'. License unchanged.

https://raw.githubusercontent.com/lightpanda-io/browser/main/LICENSE · as of 2026-10-25

Open-source microVM monitor for isolating agent code.

Limit not stated

License: Apache-2.0

Jobs: agent-sandbox, code-interpreter

LICENSE fetched 2026-10-03 is Apache License 2.0. This is the OSS runtime, not a hosted sandbox minute price. Re-fetched twice 2026-10-25 (~10 s apart, plain curl, no login): HTTP 200, no redirect, byte-identical: 'Apache License Version 2.0, January 2004'. License unchanged.

https://raw.githubusercontent.com/firecracker-microvm/firecracker/main/LICENSE · as of 2026-10-25

Open-source application kernel for sandboxing processes.

Limit not stated

License: Apache-2.0

Jobs: agent-sandbox

LICENSE fetched 2026-10-03 is Apache License 2.0. Re-fetched twice 2026-10-25 (~10 s apart, plain curl, no login): HTTP 200, no redirect, byte-identical: 'Apache License Version 2.0, January 2004'. License unchanged.

https://raw.githubusercontent.com/google/gvisor/master/LICENSE · as of 2026-10-25

Open-source unprivileged sandbox tool for Linux.

Limit not stated

License: LGPL-2.1

Jobs: agent-sandbox

COPYING fetched 2026-10-03 is GNU LGPL version 2.1. Re-fetched twice 2026-10-25 (~10 s apart, plain curl, no login): HTTP 200, no redirect, byte-identical: 'GNU LESSER GENERAL PUBLIC LICENSE Version 2.1, February 1999'. License unchanged.

https://raw.githubusercontent.com/containers/bubblewrap/main/COPYING · as of 2026-10-25

Open-source library for vector similarity search.

Limit not stated

License: MIT

Jobs: self-host-vector, embedded-vector

LICENSE fetched 2026-10-03 is MIT, copyright Facebook, Inc. and its affiliates. Re-fetched twice 2026-10-25 (~10 s apart, plain curl, no login): HTTP 200, no redirect, byte-identical: 'MIT License' / 'Copyright (c) Facebook, Inc. and its affiliates.'. License unchanged.

https://raw.githubusercontent.com/facebookresearch/faiss/main/LICENSE · as of 2026-10-25

Open-source vector database you self-host.

Limit not stated

License: Apache-2.0

Jobs: self-host-vector, billion-scale-vector

LICENSE fetched 2026-10-03 is Apache License 2.0. Hosted Zilliz Cloud is not this row. Re-fetched twice 2026-10-25 (~10 s apart, plain curl, no login): HTTP 200, no redirect, byte-identical: 'Apache License Version 2.0, January 2004'. License unchanged.

https://raw.githubusercontent.com/milvus-io/milvus/master/LICENSE · as of 2026-10-25

Open-source SQLite extension for vector search.

Limit not stated

License: MIT

Jobs: embedded-vector, edge-sqlite-agent-db, self-host-vector

LICENSE-MIT fetched 2026-10-03 is MIT, copyright 2024 Alex Garcia. A path named LICENSE.md was 404. Re-fetched twice 2026-10-25 (~10 s apart, plain curl, no login): HTTP 200, no redirect, byte-identical: 'MIT License' / 'Copyright (c) 2024 Alex Garcia'. License unchanged.

https://raw.githubusercontent.com/asg017/sqlite-vec/main/LICENSE-MIT · as of 2026-10-25

Open-source MCP server you run yourself to call GitHub. A GitHub token is still required to use the GitHub API.

Limit not stated

License: MIT

Jobs: agent-tools

LICENSE fetched 2026-10-03 is MIT, copyright 2025 GitHub. Not indexed as a hosted free API. The token is yours, not a Depot quota. Re-fetched twice 2026-10-25 (~10 s apart, plain curl, no login): HTTP 200, no redirect, byte-identical: 'MIT License' / 'Copyright (c) 2025 GitHub'. License unchanged.

https://raw.githubusercontent.com/github/github-mcp-server/main/LICENSE · as of 2026-10-25

Open-source MCP server for Chrome DevTools you run locally.

Limit not stated

License: Apache-2.0

Jobs: self-host-browser

LICENSE fetched 2026-10-03 is Apache License 2.0. Re-fetched twice 2026-10-25 (~10 s apart, plain curl, no login): HTTP 200, no redirect, byte-identical: 'Apache License Version 2.0, January 2004'. License unchanged.

https://raw.githubusercontent.com/ChromeDevTools/chrome-devtools-mcp/main/LICENSE · as of 2026-10-25

Open-source MCP server that drives Playwright locally.

Limit not stated

License: Apache-2.0

Jobs: self-host-browser, web-agent

LICENSE fetched 2026-10-03 is Apache License 2.0. Distinct from the Playwright library row already indexed as a dev tool. Re-fetched twice 2026-10-25 (~10 s apart, plain curl, no login): HTTP 200, no redirect, byte-identical: 'Apache License Version 2.0, January 2004' / 'Copyright (c) Microsoft Corporation.'. License unchanged.

https://raw.githubusercontent.com/microsoft/playwright-mcp/main/LICENSE · as of 2026-10-25

Serverless Postgres free with no time limit and no credit card: 100 projects, 100 CU-hours per project per month, 1 GB storage per project, up to 2 CU (8 GB RAM), scale to zero after 5 minutes.

{"projects":100,"cuHoursPerProjectPerMonth":100,"storageGbPerProject":1,"maxComputeCu":2}

Provider: neon

Jobs: serverless-db

Fetched 2026-10-21. Limits are per project. Paid usage: Launch $0.106/CU-hr, $0.35/GB-month.

https://neon.com/pricing · as of 2026-10-21

D1 serverless SQLite database on the Workers Free plan: 5 million rows read per day, 100,000 rows written per day, 5 GB total storage; queries error when the daily limit is hit.

{"rowsReadPerDay":5000000,"rowsWrittenPerDay":100000,"storageGbTotal":5}

Provider: cloudflare-workers

Jobs: edge-sqlite-agent-db

Fetched 2026-10-21 (page last updated Apr 21, 2026). Free limits reset daily at 00:00 UTC.

https://developers.cloudflare.com/d1/platform/pricing/ · as of 2026-10-21

Transactional email API free forever: 3,000 emails per month, at most 100 per day, 3 custom domains, 1 webhook endpoint, 30-day data retention.

{"emailsPerMonth":3000,"emailsPerDay":100,"dataRetentionDays":30}

Provider: resend

Jobs: transactional-email

Fetched 2026-10-21: 'we provide a free-forever plan'; daily limit 100 on Free; 3,000/month per the stored Free row. No overage on Free.

https://resend.com/pricing · as of 2026-10-21

100 transactional emails per month, never expires; no overages allowed on this plan. Meant for testing an integration.

{"emailsPerMonth":100}

Provider: postmark

Jobs: transactional-email

Fetched 2026-10-21.

https://postmarkapp.com/pricing · as of 2026-10-21

Hosting for personal projects at $0/month: 1M CDN requests, 100 GB Fast Data Transfer, 4 hours Fluid Active CPU, 360 GB-hrs provisioned memory, 1M function invocations per month.

{"cdnRequestsPerMonth":1000000,"fastDataTransferGbPerMonth":100,"activeCpuHoursPerMonth":4,"functionInvocationsPerMonth":1000000}

Provider: vercel

Jobs: app-hosting

Fetched 2026-10-21. Hobby is described as for 'your web app or personal project'; check Vercel's fair-use terms for commercial use (not re-read here).

https://vercel.com/pricing · as of 2026-10-21

Serverless message queue and scheduler: 1,000 messages per day, 10 active schedules, 1 MB max message size, 50 GB monthly bandwidth. Each delivery attempt, including each retry, counts as a message per the page FAQ.

{"messagesPerDay":1000,"activeSchedules":10,"maxMessageSizeMb":1,"bandwidthGbPerMonth":50}

Provider: upstash-qstash

Jobs: serverless-agent-queue, event-driven-agent-jobs

Fetched 2026-10-21. Retry billing conflict on the page's own markdown (see pack everyday-infra cron-scheduled-jobs).

https://upstash.com/pricing/qstash · as of 2026-10-21