LiteLLM

Unified OpenAI-format SDK and proxy that routes to 100+ providers.

LiteLLM is a library and proxy server that provides a unified, OpenAI-compatible interface for calling 100+ LLM providers including OpenAI, Anthropic, Azure, Bedrock, and local models. It standardizes input/output formats and adds features like retries, fallbacks, load balancing, cost tracking, and rate limiting. It is commonly used as a gateway to avoid vendor lock-in and centralize LLM access.