LiteLLM
Gateways & routersAn open-source proxy that gives 100+ model providers one OpenAI-compatible API.
LiteLLM puts one consistent, OpenAI-shaped interface in front of every major model provider. Application code calls one API; swapping providers becomes a config change, not a rewrite.
Run as a proxy server it adds the operational layer too: per-team keys, spend limits, rate limiting, fallbacks, and logging — one control point for all model traffic.
Where it's ideally used
The right layer when a system uses several model providers and you want routing, keys, and spend controlled in one place.
Where it doesn't fit
Avoidable indirection for a small project locked to a single provider with no plans to switch.