LiteLLM is an open-source Python library and gateway that lets you call models from over 100 providers through one OpenAI-style interface.
In plain words
Every model provider has its own SDK, login method, request format and error types, which gets messy when an app uses several. LiteLLM hides those differences: you write one kind of call and change only the model name. It can run inside your Python code, or as a self-hosted proxy server that a whole team shares, adding retries, fallbacks between models and cost tracking per key or user.
A quick example
Imagine an app that uses a hosted model but wants a backup. With LiteLLM, the same completion call can point at an OpenAI model, an Anthropic model or a local Ollama model just by changing the model string, and its router can fall back to another model if one fails.
Where you'll meet it
- Apps that switch between or compare model providers.
- Company-wide LLM gateways with spend tracking.
Related
Sources
- DocsBerriAI/litellm, BerriAI (GitHub) · read 29 Sept 2026
- DocsLiteLLM documentation, LiteLLM · read 29 Sept 2026
Checked by WhatIsAI against the sources above · 29 Sept 2026. Spotted an error? Report it.