LiteLLM is an open-source, self-hosted AI gateway and Python SDK that connects applications to 140+ LLM providers and 1,892 models through one OpenAI-compatible API.
LiteLLM is an open-source, self-hosted AI gateway and Python SDK that connects applications to 140+ LLM providers and 1,892 models through one OpenAI-compatible API.
More detail
It centralizes virtual keys, budgets, rate limits, spend tracking, routing, and guardrails, and can also route to MCP servers and agents.
The MIT-licensed core runs in your own cloud, on-prem, or air-gapped; LiteLLM Enterprise adds governance and support.
Best for
Platform and infrastructure teams running AI in production that need centralized key management, hard spend caps, model routing, and self-hosted or air-gapped deployment.
More detail
Python developers who want one-call access to many providers via the SDK.
Pricing
The open-source core is MIT-licensed and free to use in production; the enterprise/ directory requires a paid license.
More detail
LiteLLM adds no per-request, per-token, or percentage fee on model cost.
LiteLLM Enterprise is an annual license priced on request capacity, installation architecture, and support (
Main limitation
Teams wanting a fully managed SaaS with zero operational burden, or organizations that cannot absorb frequent security upgrades and a self-hosted Postgres/Redis stack.
LiteLLM is an open-source, self-hosted AI gateway and Python SDK that connects applications to 140+ LLM providers and 1,892 models through one OpenAI-compatible API.
It centralizes virtual keys, budgets, rate limits, spend tracking, routing, and guardrails, and can also route to MCP servers and agents.
The MIT-licensed core runs in your own cloud, on-prem, or air-gapped; LiteLLM Enterprise adds governance and support.
A practical view of what is available and where limitations apply.
Offering
Available
Detail
OpenAI-compatible unified API
Yes
One API to 140+ providers and 1,892 models; swap models without changing app code.
Spend tracking and hard budgets
Yes
Usage and spend tracked per key, user, team, org, tool, agent, and MCP; requests stop at the cap.
Rate limits and leaked-key protection
Yes
Per-team TPM/RPM limits and protection so runaway jobs or compromised keys cannot run up bills.
Routing and cost optimization
Yes
Load balancing, lowest-cost routing, Auto Routing, response and semantic caching (Redis, S3, GCS), and prompt compression.
Guardrails and audit logs
Yes
PII masking, prompt-injection detection, secret detection, moderation, and an audit log on every request.
MCP and agent gateway
Yes
MCP servers behind one fixed endpoint with per-tool access control; A2A agents, Vertex AI Agent Engine, LangGraph, Azure AI Foundry, Bedrock AgentCore, and Pydantic AI.
Python SDK
Yes
litellm package calls 100+ LLM APIs in OpenAI format with retries, fallbacks, and cost tracking.
Self-hosted and air-gapped deployment
Yes
Official Helm chart, Terraform module, Docker images; runs on your Postgres and Redis and sends no telemetry to LiteLLM (first-party claim).
SSO, SCIM, and Audit Logs (open source)
Limited
SSO, SCIM, OIDC/JWT auth, audit logs, secret managers with key rotation, and multi-region control plane are Enterprise add-ons, not the MIT core.
Managed SaaS option
No
LiteLLM is self-hosted software; buyers operate the gateway and its Postgres/Redis dependencies.
Bot trust
How confidently AI systems can verify this profile from public evidence.
—of 100
ChatGPT—/100
Claude—/100
This profile has not completed a Bot Score assessment yet.
Evidence
Where this record came from, when it was reviewed, and how confidently it can be used.
Verification
Scribble researched these facts from the linked public sources.