LibreChat: Self-Hosted Multi-Model AI Chat UI

TL;DR: LibreChat is a TypeScript, MIT-licensed AI chat interface you can self-host. It unifies OpenAI, Anthropic, AWS Bedrock, Azure OpenAI, DeepSeek, Gemini, Mistral, OpenRouter, Vertex AI, and other providers behind a single UI, with MCP integration, agents, message search, and multi-user auth.

What it is

danny-avila/LibreChat is one of the largest self-hostable ChatGPT-style interfaces on GitHub. The description packs the full feature list: agents, MCP, skills, Anthropic, AWS, OpenAI, Responses API, Azure, Groq, o1, GPT-5, Mistral, OpenRouter, Vertex AI, Gemini, Artifacts, AI model switching, message search, Code Interpreter, langchain, DALL-E-3, OpenAPI Actions, Functions, secure multi-user auth, presets. Around 43,864 stars and 9,027 forks reflect its position as the default open-source alternative for teams who want a single front-end across many model providers.

The homepage links to https://librechat.ai/, and the project has been continuously updated since February 2023.

What it does well

LibreChat is built around three practical concerns.

Provider parity. Every major hosted and self-hosted model provider has a preset in the UI. You can switch from GPT-5 to Claude to DeepSeek to a local model without leaving the chat window. The settings page exposes per-model parameters (temperature, top-p, max tokens) and lets you save presets per conversation.

Extensibility. MCP support and a built-in agent framework mean you can register tools, point them at your own servers, and have agents call them in the same conversation. OpenAPI Actions and Functions give a way to wrap arbitrary HTTP endpoints as tools.

Self-hosting posture. Secure multi-user authentication, presets per user, and a Docker-friendly deployment story make it viable for small teams. The repository includes a docker-compose.yml and a documented librechat.yaml for end-users, agents, and MCP servers.

How to try it

The quickest path uses Docker:

git clone https://github.com/danny-avila/LibreChat.git
cd LibreChat
cp .env.example .env
docker compose up -d

Once the stack is up, open http://localhost:3080. You’ll need to fill in at least one provider key in .env (for example OPENAI_API_KEY or ANTHROPIC_API_KEY) before the chat panel becomes useful. The MCP integrations live in librechat.yaml under the mcpServers key.

For a non-Docker install, the README documents Node 18+ and a MongoDB instance.

When it fits, and when it doesn’t

LibreChat is a strong default for a team that wants one chat UI across multiple providers, especially if you also want MCP tools and per-user presets. It is less attractive for a single user who only ever talks to one provider — direct provider UIs are leaner and quicker to set up. If your priority is local-only inference, look at lighter clients such as Open WebUI; LibreChat is closer to a ChatGPT Teams replacement than a private local playground.

Source and Accuracy Notes

• Project page: https://github.com/danny-avila/LibreChat
• Source repository: danny-avila/LibreChat (verified via GitHub API html_url)
• License: MIT (verified via GitHub API license.spdx_id)
• Stars at time of writing: 43,864
• Language: TypeScript
• Homepage: https://librechat.ai/