LiteLLM
Unified API for 100+ LLM providers with OpenAI-compatible interface
View on GitHubWhat it solves
Provider lock-in: writing code that works across OpenAI, Anthropic, and self-hosted models without conditional branches or SDK switching.
Overview
LiteLLM proxies calls to 100+ LLM providers (OpenAI, Anthropic, Bedrock, Vertex, Ollama) through a single OpenAI-compatible interface. Handles retries, fallbacks, budget tracking, and logging. Lets you swap providers without rewriting application code.
Key facts
- Language
- Python
- License
- MIT
- Maturity
- Community canon
- Maintainer
- BerriAI
- Reviewed
- 2026-06-11
Where to start
The proxy quickstart shows the architecture that makes provider-switching click.
docs/my_first_request.mdTechnologies
More in this domain
Instructor
Structured LLM outputs via Pydantic validation
Getting reliably structured data out of an LLM without writing validation and retry logic yourself.
Outlines
Constrained text generation using regex and context-free grammars
Guaranteed-valid structured outputs from local/open-weight models: without post-processing or retry loops.