LobeHub

LobeHub (formerly Lobe Chat) is an open source AI chat and agent platform you can self-host or deploy to Vercel. It started as a ChatGPT-style web client for talking to multiple large language model providers, and has since grown into what its maintainers call a "Chief Agent Operator" for running a team of AI agents. The interface handles chat, agent building, and multi-agent collaboration in a single deployment. Because it's self-hosted, conversations and agent configurations stay on infrastructure you control instead of a closed vendor's servers, and the project ships both English and Chinese documentation for its community.

Key features

• Multi-provider chat: Connect to OpenAI, Anthropic Claude, and other model providers from one chat interface instead of juggling separate accounts.
• Agent Builder: Describe what you need in plain language and LobeHub applies auto-configuration to set up a working agent right away.
• 10,000+ skills and MCP plugins: Extend agents with a library of tools and Model Context Protocol compatible plugins.
• Agent Groups: Assemble multiple agents for a task so they can work on it in parallel and refine results iteratively.
• Pages: Write and refine longer content with several agents in one shared workspace instead of separate chat threads.
• Personal Memory: Agents build up an understanding of how you work across sessions instead of starting from zero each conversation.
• IM Gateway: Reach agents from chat apps you already use rather than only through a dedicated dashboard.

Ideal use cases

LobeHub fits individuals who want a private, self-hosted chat client instead of a closed AI app, and teams that want to run multiple AI agents against shared context without depending on a single vendor's hosted platform. It also suits developers who want one interface in front of several model providers, and organizations that want a Workspace for tracking ownership of agent-driven work across a team rather than scattering it across individual chat histories. Teams evaluating a self-hosted AI operations layer, rather than a single chat window, are the clearest fit, since Agent Groups and Pages assume more than one agent is working on a task at a time.

Self-hosting and deployment

LobeHub deploys with the official Docker image, or with one-click templates for Vercel, Zeabur, Sealos, and Alibaba Cloud. Configuration is handled through environment variables for model provider API keys and other settings, so switching between OpenAI, Anthropic, and other providers doesn't require code changes. The project also documents a local development setup for contributors building from source, along with a Docker Compose path for running the full stack on your own server. Because the project is under active development, the maintainers recommend starring the repository and watching release notes so self-hosted instances don't fall behind new features like Agent Builder or the IM Gateway.