Compare GPU rental prices across neo cloud providers — updated nightly.
A real-time dashboard for developers, AI engineers, and ML researchers who need to find the cheapest H100, A100, L40S, or B200 GPU instances for training, fine-tuning, or inference workloads. Browse pricing from AWS, Azure, GCP, Oracle, RunPod, Lambda Labs, Nebius, Crusoe and more in one place — no vendor lock-in, no sign-up required.
Live: gpuprice.becloudready.com
- Nightly price scraping
- Real-time comparison dashboard
- Historical price charts
- Multi-GPU comparison
- Filters
Prices come from machine-readable sources, not HTML scraping. Each source is an independent job: if one fails, its last good data is kept and flagged stale in meta.json.
| Source | Providers |
|---|---|
| dstack gpuhunt catalog | AWS, Azure, GCP, Oracle, Lambda, RunPod, Nebius, Verda, CloudRift |
| Shadeform marketplace API | Crusoe, Denvr, Massed Compute, Paperspace, Hyperstack, Scaleway, Latitude and others |
| Akamai/Linode API | Linode |
| Vultr API | Vultr |
All prices are on-demand USD per GPU-hour (spot, reserved and fractional-GPU offers are excluded). Rows outside per-model sanity bounds are quarantined, not published.
scraper/ Python scrapers + merge step (run nightly by GitHub Actions)
sources/ one module per upstream source
schema.py normalized Offer schema, GPU naming, sanity bounds
run_all.py runs each job independently, with stale fallback
merge.py merges job files into all.json, all.csv, meta.json
tests/ pytest suite for the scraper
nextjs-dashboard/ Next.js frontend, deployed to Cloudflare Workers via OpenNext
.github/workflows/ nightly scrape and upload to R2
pip install -r scraper/requirements.txt
python -m scraper.run_all --out-dir data --raw-dir raw
python -m scraper.merge --data-dir data
pip install -r scraper/requirements-dev.txt && python -m pytest testscd nextjs-dashboard
npm install
npm run devOpen http://localhost:3000.
Apache 2.0