LiteLLM
An OpenAI-compatible gateway for routing, governing, and tracking LLM access across many providers.
About LiteLLM
LiteLLM is an OpenAI-compatible LLM gateway built for teams that want one control point across many model providers. Applications keep a familiar request format while the gateway manages authentication, model access, fallbacks, budgets, rate limits, spend tracking, and logging. It fits best when several applications or teams share model infrastructure and provider-specific integrations have become hard to govern. The open-source edition includes the core gateway and operational controls. Enterprise adds SSO, SCIM, delegated administration, stronger secret handling, multi-region deployment, and professional support. Self-hosting gives teams control, but it also leaves them responsible for deployment, configuration, monitoring, and upgrades. Enterprise pricing has no posted fixed amount, so larger organizations must request a quote and weigh the added controls against their deployment and support needs.
Best for
Best use cases
An application is adding a second model provider while keeping its existing OpenAI-style request format.
Several internal products share model access and need per-key or per-team spend limits and reporting.
A production service needs fallbacks, guardrails, monitoring, and observability integrations around model calls.
Compare before choosing
Compared with direct provider integrations, LiteLLM adds a gateway layer but reduces provider-specific client and policy work.
The free OSS edition covers core routing and controls, while Enterprise reserves some large-organization features for a quote-based license.
Self-hosting provides infrastructure control, but a managed gateway may require less deployment, monitoring, and upgrade work.
Features
Unified model gateway
Route application traffic to supported LLM providers through one OpenAI-compatible gateway and reduce provider-specific client code.
Cost and access control
Issue virtual keys, track spend, set budgets and rate limits, and restrict model access for keys, users, or teams.
Resilient production routing
Configure fallbacks, monitoring, logging, guardrails, and observability integrations for production model traffic.
Pros
- One OpenAI-compatible interface can reduce provider-specific integration work across applications.
- The free OSS edition includes virtual keys, spend tracking, budgets, fallbacks, and request logging.
- Centralized model access, rate limits, and observability give platform teams practical operational control.
Cons
- Self-hosted deployments require engineering time for configuration, monitoring, maintenance, and upgrades.
- Enterprise pricing is quote-based, which makes early cost comparison less direct for larger organizations.
- Some controls for large organizations, including broader SSO and administration features, require Enterprise.
Pricing Plans
OSS
- OpenAI-compatible gateway
- Virtual keys and authentication
- Spend tracking, budgets, fallbacks, and request logging
Enterprise
- SSO and SCIM controls
- Delegated administration and secret-manager support
- Multi-region deployment and professional support
* Pricing information is for reference only. Please refer to the latest announcement on LiteLLM's official website.