Introduction
Real-time market data for GPU compute, over one REST API.
The OpenGrid Data API gives you the GPU compute market as data: every price OpenGrid observes across cloud and marketplace providers, the indices and fixings built from them, capacity, the full cost of a job, provider reliability, market events and news.
https://data.tryopengrid.com/v1Every request needs an API key. Every response says how fresh its data is.
Quickstart
Make your first request in two minutes.
API reference
Every endpoint, its fields, and a playground to call it with your key.
Authentication
How to get a key and send it.
For AI agents
llms.txt, markdown pages and the OpenAPI schema.
What you can get
| Area | What it answers | Start with |
|---|---|---|
| Markets | What does an H100 cost right now, across every provider? | GET /v1/markets |
| Listings | Which exact offers exist, where, and are they in stock? | GET /v1/listings |
| History | How has that price moved over 24 hours, 90 days, or as far back as archives go? | GET /v1/markets/{gpu}/history |
| Indices and fixings | What is the reference price of a GPU, and can I verify it? | GET /v1/indices, GET /v1/fixings |
| Capacity | What do providers report as available? | GET /v1/capacity |
| Costs | What does a whole job cost, with egress and storage? | POST /v1/costs/estimate |
| Reliability | Is a provider's API up, and how fresh is its data? | GET /v1/reliability/providers |
| Events and news | What changed in the market, and what was reported? | GET /v1/events, GET /v1/news |
How it works
OpenGrid polls each provider's own API or price list every 15 to 60 minutes. It maps every offer to a canonical GPU, converts it to USD per GPU-hour, and checks it before publishing. Every row carries when it was observed, and corrections are never silent.
Access
The Data API is available to companies under an enterprise agreement. To get a key, contact us at hello@tryopengrid.com.