Overview

LiteLLM is the industry's fastest and lightest AI Gateway, empowering developers to seamlessly interact with over 100 Large Language Models (LLMs) through a single, unified API. Built with a robust Rust core and a user-friendly Python SDK, LiteLLM simplifies LLM integration by offering OpenAI-compatible endpoints. This means you can call models from providers like OpenAI, Anthropic, Bedrock, Azure, and Vertex AI with minimal code changes, all while benefiting from advanced features like cost tracking, intelligent guardrails, efficient load balancing, and comprehensive logging. Deploy LiteLLM as a self-hosted, enterprise-ready proxy server or integrate its powerful Python SDK directly into your applications.

Key Features

• Universal LLM Access: Integrate with 100+ LLM providers, including major players and emerging models, using a single API format.
• OpenAI Compatibility: Effortlessly switch between LLM providers without modifying your existing codebase, thanks to drop-in OpenAI format support.
• High-Performance Gateway: Experience lightning-fast response times with a Rust-based core, achieving P95 latency as low as 8ms at 1k RPS.
• Production-Ready Features: Benefit from enterprise-grade capabilities such as virtual API key management, real-time cost tracking, customizable guardrails, intelligent load balancing, and detailed logging.
• Flexible Integration: Choose between a powerful Python SDK for direct integration or deploy the AI Gateway (Proxy Server) as a centralized service for your team.
• Agent and MCP Support: Extend LLM capabilities with integrated support for Agentic workflows (A2A Agents, Agent Harnesses) and Meta-Model Communication Protocol (MCP) tools.
• Cost Management: Gain clear visibility into your LLM spending with built-in cost tracking across all integrated providers.

Typical Use Cases

• Unified LLM Application Development: Build applications that can dynamically switch between different LLM providers based on performance, cost, or specific model capabilities without code refactoring.
• Centralized LLM Management for Teams: Deploy the LiteLLM Gateway as a central hub for an organization's LLM usage, enforcing policies, tracking costs, and simplifying API key management.
• Rapid Prototyping and Experimentation: Quickly test and compare different LLMs for your use case by simply changing the model parameter in your requests.
• Building AI-Powered Tools and Platforms: Integrate LLM capabilities into developer tools, IDEs (like Cursor), and other platforms with a consistent and reliable API.
• Implementing Advanced AI Workflows: Leverage LiteLLM's support for Agents and MCP tools to create sophisticated, multi-model AI systems and connect with external services.