LobeChat started as an open-source chat UI for language models and has grown into a full assistant workspace now branded LobeHub. The interface supports OpenAI, Anthropic Claude, Google Gemini, Mistral, DeepSeek, Qwen, Groq, Ollama and other providers, and you can switch between them mid-conversation or run local models through Ollama. Conversations are organised into assistants and topics, with prompt libraries, branch-and-edit message trees, vision input for image understanding and function calling for tools.
The product layer around the chat is what distinguishes it: an assistant marketplace for discovering and sharing agents, a knowledge base that uploads documents and grounds answers in your own files, and a plugin system that lets assistants search the web, generate images, read PDFs or call external APIs. Artifacts render generated code and documents in a live canvas, and text-to-speech adds voice playback. Because it is open source, every part of the experience — including the model gateway and data storage — can be deployed on your own infrastructure.
LobeChat is used as both a free personal assistant and a team tool. Individual users get a capable daily chat client without vendor lock-in, while organisations that care about data residency can self-host the whole stack and point it at internal or Azure OpenAI endpoints.
