Everything you need to build on Inferray
Inferray is one API for 21 flagship models from Anthropic, OpenAI, Google, DeepSeek, Z.ai, and Moonshot AI. It speaks the request formats your code already uses, so integrating is a base URL and a key rather than a rewrite.
Base URL
https://api.inferray.com/v1
Authentication
Authorization: Bearer inferray_...
Start here
Three pages get you from nothing to a billed request.
The APIs
Four request formats, one catalogue of models behind all of them. Full reference pages are on the way; for now each links to the worked example in the quickstart.
- Chat Completions/v1/chat/completionsThe OpenAI-compatible endpoint. Point the openai SDK at Inferray and change nothing else.
- Responses/v1/responsesThe other OpenAI-compatible endpoint, and the one Codex speaks. Streaming and store are passed through as sent.
- Messages/v1/messagesThe Anthropic-compatible endpoint, served for the whole catalogue rather than only Claude. Takes x-api-key or a bearer token.
- Image generation/v1/images/generationsThe OpenAI image shape, for the models that produce images. Priced per image or per request rather than per 1M tokens.
- Models/v1/modelsThe catalogue in the shape OpenAI clients expect, so model pickers populate themselves. Free, but authenticated.
Guides
The things you reach for once the first request works.
Ready to send a request?
Create a key, top up, and point your existing client at Inferray.