← Back to stack

Builder infrastructure

LiteLLM

The AI Gateway for platform teams.

Visit official site →

LiteLLM positions itself as a self-hostable AI gateway that puts model access, routing, spend controls, and governance behind one OpenAI-compatible interface.

Vendor's own claim

What LiteLLM is

Vendor's own claim

LiteLLM offers a Proxy Server for a central LLM gateway and a Python SDK for direct application use.

What it can do

Vendor's own claim
  • OpenAI-compatible access to 100+ providers
  • Model routing and fallbacks
  • Spend tracking, budgets, and rate limits
  • Self-hosted and air-gapped deployment

Who it is for

Vendor's own claim

Platform teams standardizing access to multiple LLM providers.

When to choose something else

Our read

A small application using one provider may not need a gateway's routing, governance, virtual-key, and budget configuration surface.

Implementation considerations

Our read

Setup assessment: 2h

Our read: the vendor says self-hosting takes minutes, while a production gateway requires provider and governance configuration.

Developer and agent access

From official docs
Base URL
https://models.litellm.ai
Machine-readable API
No OpenAPI link in the reviewed source set
  • GET /model_catalog
  • GET /model_catalog/{model}
  • POST /billing/checkout
  • GET /billing/usage

Published pricing

From official docs
PlanPriceIncluded
Open Source$0Free forever · self-hosted
EnterpriseTalk to salesAnnual · sized to your usage

Verify before you adopt it

  • No confirmed MCP-specific documentation or tool list.
  • A machine-readable OpenAPI specification was not included in the reviewed source set.