The modern open-source AI chat framework — multiple providers, knowledge bases, multi-modal input and one-click plugins, deployable free in minutes. A flexible alternative to hosted assistants.
LobeChat is the design-forward take on self-hosted AI chat: a gorgeous interface backed by a real framework — dozens of model providers, a plugin/agent marketplace, knowledge bases with file upload, and multi-modal chat with vision, voice and image generation.
Its agent marketplace is the standout: hundreds of ready-made assistants and one-click plugins (search, crawlers, generators) that extend chats without configuration. Branching conversations and artifacts support serious workflows.
Apache-2.0 licensed and deployable free on Vercel or with Docker in minutes, LobeChat is the most polished way to run your own multi-provider AI assistant.
Yes — open source under Apache-2.0. Deploy it free on Vercel or Docker; you only pay for the model APIs you use, or nothing with local models.
Dozens — OpenAI, Anthropic Claude, Google Gemini, Ollama local models, DeepSeek, Qwen and many more, all switchable per conversation.
Both are excellent; LobeChat leans design-polish and a rich agent/plugin marketplace, while Open WebUI leans local-first depth and admin controls. Many people run both.
Yes — upload files and chat over them with retrieval, alongside vision and voice input.
One-click Vercel deployment for a personal copy, or Docker Compose with Postgres for the full server version.