Predictable dedicated GPU capacity
Labs need known capacity and clusters without marketplace chaos.
At a glance — Oct 2026
Updated 2026-10-27 · forward this page: /jtbds/predictable-gpu#briefScores as of 2026-09-14 (43 days old; Depot marks scores stale after 30 days). The pick is still Depot's current editorial pick; it has not been re-reviewed since. Prices below carry their own dates.
- Depot Pick
- Lambda · 7.82/10 for this job (curated estimate, high confidence)
reliability + maturity + capability (interconnect).
- Runner-up
- CoreWeave · 7.47/10
Scores higher than the pick on: Capability fit, Latency / reliability, Privacy / data retention, Maturity / track record.
- When not to use the pick
- Need max enterprise K8s (CoreWeave) or burst serverless (Modal/RunPod).
- Lock-in / migration
- The pick has lower stored lock-in risk than runner-up CoreWeave (5.5 vs 6.5).
- Sourced price (usd_per_gpu_hour)
Pick's lowest sourced SKU: NVIDIA A100 SXM 80GB $2.79 (usd_per_gpu_hour) — source, as of 2026-10-19
Cheapest matching SKU: RunPod RTX 4090 Pod (Community Cloud) $0.34 — source, as of 2026-10-19. Not the pick; not quality-adjusted.
Price history: first seen 2026-10-19 · last changed no change recorded · history JSON
- Free option
- No free resource lists this job. A cheap paid SKU is not a free option.
- Failure modes to watch
- Runaway spend: Set a budget and a tool-call cap per task. Alert when spend jumps. Stop the run at the cap instead of summarizing a partial trace as success.
Same-job, same-unit comparisons: Lambda vs CoreWeave · Lambda vs Crusoe · Lambda vs RunPod
Cite: Agent Provider Depot, Depot Pick for predictable-gpu: lambda, scores as of 2026-09-14. Agents: GET /api/decide?jtbd=predictable-gpu
All sourced prices, cheapest first
Lambda
reliability + maturity + capability (interconnect).
Pricing (directional): On-demand H100 SXM ~$3.99–4.29/hr; 1-Click Clusters for multi-week reservations; no classic spot tier.
Lock-in risk: higher = more lock-in (worse). Overall already inverts this dimension.
Best boring GPU cloud for research labs who want clusters without CoreWeave enterprise theater.
Quote for Predictable dedicated GPU capacity
Depot Pick stays Lambda. Prices below are sourced SKU rows only (as of 2026-10-27). Input and output are not blended. GET /api/quote?jtbd=predictable-gpu
Pick is not a price winner. lambda matching USD per GPU-hour is 2.79; cheapest among candidates is runpod at 0.34.
- NVIDIA A100 SXM 80GB usd_per_gpu_hour $2.79 source as of 2026-10-19
- NVIDIA B200 SXM6 usd_per_gpu_hour $6.69 source as of 2026-10-19
- NVIDIA H100 SXM usd_per_gpu_hour $3.99 source as of 2026-10-19
6 of 13 candidates have no sourced price row. 6 are excluded from the USD per GPU-hour comparison (gpu_hour). min sourced GPU-hour SKU among candidates; not coerced to $/1M; per-second GPU was not included; not quality-adjusted
Sourced USD per GPU-hour for Predictable dedicated GPU capacity
Candidates for this job that already have a matching sourced price (gpu_hour), cheapest first. As of 2026-10-27. Each figure is one SKU in that unit, not a blend of input and output, not a cross-modal rate, and not a Depot score. min sourced GPU-hour SKU among candidates; not coerced to $/1M; per-second GPU was not included; not quality-adjusted The Depot Pick is not a price winner.
| Provider | USD per GPU-hour | SKU | Source | As of |
|---|---|---|---|---|
| RunPod | $0.34usd_per_gpu_hour | RTX 4090 Pod (Community Cloud) | pricing page | 2026-10-19 |
| Massed Compute | $0.97usd_per_gpu_hour | L40S (48GB) x1 | pricing page | 2026-10-19 |
| Crusoe | $1.5usd_per_gpu_hour | NVIDIA L40S 48GB | pricing page | 2026-10-19 |
| Nebius | $1.8usd_per_gpu_hour | NVIDIA RTX PRO 6000 | pricing page | 2026-10-19 |
| Verda | $1.845usd_per_gpu_hour | 1x L40S 48GB | pricing page | 2026-10-27 |
| LambdaDepot Pick | $2.79usd_per_gpu_hour | NVIDIA A100 SXM 80GB | pricing page | 2026-10-19 |
| CoreWeave | $6.16usd_per_gpu_hour | NVIDIA HGX H100 inference single GPU | pricing page | 2026-10-19 |
Runners-up
CoreWeave
Enterprise training/inference fleet pick. Skip if you just need a single H100 for an afternoon — use RunPod.
Crusoe
Solid neocloud when you want SLA + energy story. Not the first stop for hackathon GPUs.
RunPod
Depot default for self-serve GPUs. Secure Cloud for anything customer-facing; Community/spot for batch.
Verda
Top EU list-price neocloud peer to Nebius. Prefer Nebius Depot default for global neocloud; Verda when EU…
Massed Compute
Strong self-serve neocloud with a durable public rate card and owned fleet (not a marketplace). Good…
Hyperstack
Solid UK/EU GPU alternative with public card. Nebius/Verda first for neocloud; Hyperstack when UK entity +…
When NOT to use the top pick
Need max enterprise K8s (CoreWeave) or burst serverless (Modal/RunPod).
Full ranking (score table)
Ranked with this JTBD's weights. Scores are curated estimates unless marked otherwise. Agents: prefer /api/decide then this table for tie-breaks.
| # | Provider | Overall | Fit | $/perf | Latency | DX | As of |
|---|---|---|---|---|---|---|---|
| 1 | RunPod | 8.3 | 8.8 | 9.0 | 8.0 | 8.8 | 2026-10-19 |
| 2 | Verda | 8.2 | 8.6 | 8.2 | 8.4 | 8.4 | 2026-09-14 |
| 3 | Koyeb | 8.1 | 8.2 | 8.4 | 8.2 | 8.8 | 2026-09-14 |
| 4 | Gcore | 8.1 | 8.0 | 8.8 | 8.2 | 7.2 | 2026-09-14 |
| 5 | Hyperstack | 8.0 | 8.2 | 8.4 | 8.0 | 8.0 | 2026-09-14 |
| 6 | LambdaDepot Pick | 7.8 | 8.8 | 6.8 | 8.8 | 8.5 | 2026-09-14 |
| 7 | Crusoe | 7.7 | 8.6 | 7.2 | 8.6 | 7.4 | 2026-09-14 |
| 8 | Nebius | 7.7 | 8.6 | 7.0 | 8.4 | 7.6 | 2026-09-14 |
| 9 | CoreWeave | 7.5 | 9.2 | 5.5 | 9.0 | 7.5 | 2026-10-19 |
| 10 | Massed Compute | 7.5 | 7.6 | 8.2 | 7.4 | 7.2 | 2026-09-14 |
| 11 | Oblivus | 7.3 | 7.4 | 8.4 | 7.0 | 7.2 | 2026-09-14 |
| 12 | CloudFerro | 7.2 | 7.2 | 5.8 | 8.4 | 6.8 | 2026-09-14 |
| 13 | Shadeform | 7.2 | 7.0 | 8.4 | 6.8 | 8.0 | 2026-09-14 |
Score methodology for this job
reliability + maturity + capability (interconnect).
capabilityFit: 22%pricePerformance: 25%latencyReliability: 15%developerDx: 12%privacyPosture: 8%lockInRisk: 8%maturity: 10%