Spot price history for AWS, GCP and Azure, from price-change events per zone to monthly series back to 2014, GPU rental prices across 16 providers with daily index fixings, and LLM token prices per model across hosts. The charts are free. Every number behind them is one HTTP call away. No account needed: pay per call over x402 or MPP, or use a prepaid key.
Spot, on-demand and reserved prices for every instance type across 30 regions, charted with indicators and exportable. Daily detail for 90 days, weekly for a year, monthly back to 2014.
Open the charts →H100, H200, B200, A100 and more in USD per GPU-hour across neo-clouds, marketplaces and hyperscaler spot, with the day's index fixing per segment.
See today's prices →Event-level spot history per zone, list prices with 1y and 3y terms, GPU index fixings for 367 tickers, token prices for 934 models, a term structure per GPU and a capacity stress score. Every fixing carries a methodology version.
Read the docs →A paid endpoint answers 402 with its price and its payment offers: USDC over x402 on 13 networks, or MPP on Tempo. The client pays and repeats the call; a prepaid key skips the handshake. The OpenAPI document carries the price of every operation, and the hosted MCP server gives any agent 35 tools, paid from a prepaid key or, in the npm package, from a wallet. How per-request payment works →
npx -y corpus-cloud-pricing-mcp
circle services pay "https://cloud.trycorpus.ai/v1/index/fixing?ticker=CX.H100SXM.NEO" -X GETSources are named on every row: provider APIs for spot and list prices, public price feeds and pages for GPU rentals, the OpenRouter, LiteLLM and DeepInfra catalogs for token prices, and the USC/ISI and SpotLake archives for the years before our own collection. The methodology explains every fixing. Built and run by corpusAI.