# Hundredth Run Compute Buyer Desk > Free, independent GPU rental price data. Every row has a source URL and a date checked. Unknown prices stay blank. Data is CC BY 4.0: credit Hundredth Run. Last data check: Nov 16, 2026. Providers tracked: 26. Price rows: 220. ## Data - [Prices as JSON](https://hundredthrun-compute-desk.pages.dev/api/v1/prices.json): every row with source and date - [GPU summaries as JSON](https://hundredthrun-compute-desk.pages.dev/api/v1/gpus.json): lowest, median and priciest public on-demand price per GPU - [Model memory estimates as JSON](https://hundredthrun-compute-desk.pages.dev/api/v1/models.json) - [OpenAPI spec for the JSON API](https://hundredthrun-compute-desk.pages.dev/api/v1/openapi.json) - [APIs.json discovery file](https://hundredthrun-compute-desk.pages.dev/apis.json) - [Remote MCP server](https://hundredthrun-compute-desk.pages.dev/mcp): streamable HTTP, no key, tools list_gpus, cheapest_gpu, get_prices (official MCP Registry name: io.github.hundredthrun/gpu-cloud-prices) - [Prices as CSV](https://hundredthrun-compute-desk.pages.dev/data/gpu_prices.csv) - [Docs and licence](https://hundredthrun-compute-desk.pages.dev/open-data) ## Pages - [Price by GPU](https://hundredthrun-compute-desk.pages.dev/gpu/): cheapest, median and priciest on-demand price per GPU - [Cheapest GPUs](https://hundredthrun-compute-desk.pages.dev/cheapest/): high/medium on-demand listings ranked cheapest first - [Spot and preemptible GPUs](https://hundredthrun-compute-desk.pages.dev/spot/): high/medium interruptible listings ranked cheapest first - [GPU price history](https://hundredthrun-compute-desk.pages.dev/history/): daily low, median and high per GPU since tracking started, with provider price changes - [Calculator](https://hundredthrun-compute-desk.pages.dev/calculator): VRAM needed and cheapest public rental for an open model - [Reserved vs on-demand](https://hundredthrun-compute-desk.pages.dev/calculator/reserved): compare sourced reserved and on-demand rates by hours per month - [Self-host vs API](https://hundredthrun-compute-desk.pages.dev/calculator/self-host): break-even between sourced GPU rental and sourced inference API $/1M token rates - [Model guides](https://hundredthrun-compute-desk.pages.dev/models/): VRAM and rental cost for 19 open models - [GPU vs GPU](https://hundredthrun-compute-desk.pages.dev/compare/): price and memory comparisons - [Provider pages](https://hundredthrun-compute-desk.pages.dev/provider/): one page per provider - [Provider vs provider](https://hundredthrun-compute-desk.pages.dev/vs/): head-to-head on-demand prices for pairs of GPU clouds, like RunPod vs Lambda - [Provider vetting checklist](https://hundredthrun-compute-desk.pages.dev/GPU-Provider-Vetting-Checklist.pdf): 24 checks before you pay a GPU provider