Introduction
infere.dev serves open-weight language models over an API that uses the OpenAI request format.
What you get
- Open-weight models hosted on our infrastructure: DeepSeek, Qwen, Llama and Gemma.
- One endpoint for every model. Change the model id to switch.
- Server-sent event streaming for chat completions.
- Pay for the tokens you use, from a prepaid credit balance.
Base URL
All API requests go to https://api.infere.dev/v1. Requests must use HTTPS. Plain HTTP is refused.
Where to start
New here? Follow the quickstart to send your first request in under five minutes. Then read authentication before you put a key in production.
Availability: the console, sign-in, API keys and billing are live. The inference endpoint is in controlled rollout; if a request returns 503 with service_unavailable, wait and retry with backoff. See the errors page.