LIVE
OPUS 4.7$15 / $75per Mtok
SONNET 4.6$3 / $15per Mtok
GPT-5.5$10 / $30per Mtok
GEMINI 3.1$3.50 / $10.50per Mtok
SWE-BENCHleader Claude Opus 4.772.1%
MMLU-PROleader Opus 4.788.4
VALS FINANCEleader Opus 4.764.4%
AFTAv1.0 whitepaper live at /whitepaper
OPUS 4.7$15 / $75per Mtok
SONNET 4.6$3 / $15per Mtok
GPT-5.5$10 / $30per Mtok
GEMINI 3.1$3.50 / $10.50per Mtok
SWE-BENCHleader Claude Opus 4.772.1%
MMLU-PROleader Opus 4.788.4
VALS FINANCEleader Opus 4.764.4%
AFTAv1.0 whitepaper live at /whitepaper
All systems operational0 AI providers monitored, polled every 2 minutes
Live status

GPU Rental Pricing

Live cheapest hourly rates across the cloud GPU marketplaces. Refreshed every 4 hours.

GPU prices move. A100s that cost $4 an hour on a hyperscaler can be $0.80 on a marketplace in the same week, and the cheapest provider for an H100 today might be the most expensive tomorrow. This page aggregates current per-GPU hourly rates across marketplace providers, normalizes their heterogeneous GPU naming into a canonical taxonomy, and surfaces the cheapest on-demand and spot price for each GPU class.

Sources: RunPod (when an API key is configured) plus a manual snapshot of Lambda Labs on-demand pricing from lambda.ai/pricing. CoreWeave public pricing, Azure NC/ND, and AWS on-demand are planned next, contingent on per-source ToS review. The data is captured daily into a historical snapshot. The 30 to 90 day price series is exposed via /api/gpu/pricing/series, free and agent-friendly.

Free agent endpoints

  • /api/gpu/pricingFull current snapshot
  • /api/gpu/pricing/cheapest?gpu=H100&type=on_demandTop 3 cheapest right now
  • /api/gpu/pricing/series?gpu=H100&from=&to=Daily price series, up to 90 days. Backfill is impossible; every day matters.

Frequently asked questions

Where does TensorFeed get GPU pricing data?
Phase 1 sources are Vast.ai (public marketplace API) and RunPod (GraphQL API). Lambda Labs, CoreWeave, Azure, and AWS are planned for phase 2.
How often is the data refreshed?
Every 4 hours. A daily snapshot is also captured at 12:45 UTC for the historical price series exposed via the premium API.
What is the difference between on-demand and spot pricing?
On-demand is the uninterruptible hourly rate. Spot (or interruptible / bid) is a lower rate where the provider can reclaim the machine on short notice. Spot is great for fault-tolerant batch jobs, not for serving production traffic.
Can I get a programmatic feed?
Yes. /api/gpu/pricing returns the full snapshot. /api/gpu/pricing/cheapest?gpu=H100 returns the top 3 cheapest right now. /api/gpu/pricing/series returns the daily historical price series for a canonical GPU. All three are free, no auth required.