danny-avila/LibreChat

A self-hosted conversational AI platform that lets you connect virtually every major model provider through a single, unified interface. Instead of juggling separate logins for OpenAI, Anthropic, Gemini, Mistral, Groq, and others, you get one chat frontend that routes to whichever model you choose per conversation. On top of raw model access, it layers in agents, tool calling via MCP, a built-in code interpreter, DALL-E image generation, and OpenAPI action integrations, so it behaves less like a simple chat wrapper and more like a full-featured AI workspace. The multi-user authentication system with role-based access makes this genuinely practical for small teams or internal company deployments where you want to share infrastructure without sharing credentials or conversation history. Setup requires Docker familiarity and some configuration work, which is the honest barrier here. It is not a five-minute install if you want everything wired correctly. -> Best for: technical founders or platform teams building an internal AI hub who want full control over data, costs, and model routing without writing a frontend from scratch.