InferBridge Docs

One API for every LLM — global, open-source, and Indian.

InferBridge is a drop-in, OpenAI-compatible gateway. Point your existing OpenAI SDK at https://inferbridge.dev/v1, keep your prompts and streaming code unchanged, and get tier-based routing, per-request caching, residency filtering, and per-call observability across OpenAI, Anthropic, Together, Sarvam, and your own self-hosted endpoints.

What’s here

Getting started — thirty seconds from curl to first response.
Migrating from OpenAI — the two-line patch for Python and Node.
API reference — every endpoint, every header, every response field.
Changelog — releases, breaking changes, and deprecation windows.

Not yet covered here

Some surface area deliberately lives outside /docs:

Architecture + internal design — see the main repo README and the docs/design/ folder in-repo.
Deploy + ops runbooks — the backend repo’s docs/deploy/ folder.
Provider-specific quirks and pricing tables — consolidated in /docs/api/chat-completions/.

InferBridge was released as Agni AI in v0.1.0; the rename shipped in v0.2.0 on 2026-04-23. Legacy agni_* keys, agni/* modes, and X-Agni-* headers remain accepted until 2026-07-22. See the changelog for the full compatibility matrix.