LibreChat

LibreChat is an open-source, self-hosted web app and server that provides a unified web interface that allows users to interact with multiple AI providers and local models through a single platform. The software is deployed as a server application, often using Docker, and supports both local and cloud installations for users who want to maintain control over their AI infrastructure.

The platform acts as a centralized hub for various large language models, enabling users to switch between different AI endpoints mid-conversation. It includes tools for managing multi-user access, conversation history, and custom AI agents. Users can deploy the system on a single server or scale it horizontally using Redis for production environments to ensure stability across multiple devices and tabs.

Key features

• Integration with OpenAI, Anthropic, Google, AWS Bedrock, and Azure OpenAI
• Support for local providers including Ollama, Mistral AI, and Deepseek
• Sandboxed code interpreter for executing Python, Node.js, Go, and other languages
• No-code agent builder with a community marketplace and MCP server support
• Generative UI artifacts for creating React, HTML, and Mermaid diagrams
• Multimodal capabilities for image analysis and text-to-image generation
• Multi-user authentication via OAuth2, LDAP, and email login
• Admin panel for managing users, roles, and configuration overrides
• Web search integration with customizable reranking services

LibreChat is designed for individuals and organizations that require strict control over their AI infrastructure and data privacy. It supports a wide range of integrations, including S3 with CloudFront for stable media links and various speech-to-text and text-to-speech providers like Elevenlabs. The architecture allows for the creation of reusable instruction bundles called skills and the delegation of focused work to isolated subagents with their own context windows. It also includes a reasoning UI specifically for chain-of-thought models like DeepSeek-R1.

This software serves as a comprehensive AI orchestration layer for those who use multiple model providers and need a centralized, privacy-focused management system.