Choose your model.

Models, one API.
Inference API · separate input and output units

Works with the OpenAI SDKs

When inference is available, point the official SDK at the base URL and use your API key. The supported endpoints are chat/completions, completions, embeddings and models.

Prices in reais

For models priced in US dollars upstream, the price in reais follows the exchange rate in use, which each price shows when you open it.

Build what comes next.

Request an invitation and tell the team what you want to build.