LiteLLM

Hybrid (offline or cloud)

Drop-in OpenAI-compatible proxy across 100+ providers. Route to local Ollama or cloud, same code.

Editorial verdict: “Best universal LLM proxy. Foundational layer for multi-provider deployments.

SDK / proxy
Free tier
MIT
4.5 / 5
GitHub ★ 15,000

Compatibility at a glance

Which runtime + OS combos this app works against. Source of truth for "will it run on my setup?"

§ Runtimes supported
ollamallama-cppopenai-compatanthropicgeminiopenai
§ OS / platform
linuxmacoswindows

What it is

LiteLLM is the right choice when you need to swap between local Ollama or llama.cpp and cloud providers like Anthropic or Gemini without rewriting your app code. It exposes a single OpenAI-compatible API shape and routes to over 100 backends, making it ideal for teams A/B testing models or migrating from one provider to another. The built-in fallback, retry, and cost tracking handle operational overhead, but the free tier covers most needs—advanced features require the paid cloud version. This adds a proxy layer to your stack, so you trade simplicity for flexibility. Works on Linux, macOS, and Windows.

✓ Strengths

  • +Genuinely universal — 100+ providers
  • +OpenAI-shaped API stays put
  • +Built-in fallback / retry / cost tracking

△ Caveats

  • Some advanced features paid (LiteLLM Cloud)
  • Extra moving part in your stack