LobeChat Review

LobeHub is a platform for creating, scheduling, and running AI agents for long-running tasks, team workflows, and automated jobs. The README lists an Agent Builder that configures an agent's name, role, skills, and behavior from a description, Agent Groups that assemble and run tasks together, Pages, Schedules, Projects, and a shared team Workspace, an IM Gateway that connects agents to messaging platforms, and a Personal Memory with Continual Learning that users can inspect and edit. It works with OpenAI and Anthropic models among others, connects to a library of 10,000+ tools and MCP-compatible plugins (the README says the plugin system is undergoing major development), and OpenAlternative cites a skills library of over 312,000 entries. Releases are frequent (v2.2.19 at the time of review).

The README offers one-click deploys to Vercel, Zeabur, Sealos, RepoCloud, and Alibaba Cloud, a Docker route (run a setup script, then docker compose up -d), and local development with pnpm install and pnpm dev. The official self-hosting documentation page we tried was an unfinished placeholder, listing only the topics it will cover: deployment options (Vercel, Docker, and server database mode), required services (a Postgres database, S3-compatible storage, and an auth provider), environment variables, the difference between client-side and server-side database modes, and upgrade notes. For the default setup OPENAI_API_KEY is required, OPENAI_PROXY_URL optionally overrides the API base URL, and OPENAI_MODEL_LIST lets you add, hide, or rename models. The README recommends third-party OpenAI proxies while disclaiming their validity and security, so use your own key where possible.

The interface is a modern chat client with model switching and an agent layer on top: you describe an agent, assign it tasks, and let groups run while reporting progress. Memory is structured and editable, which is a differentiator against opaque memory in hosted assistants. The README describes the project as under active development and the plugin system as still being built in phases. This section reflects the README, license, and OpenAlternative; we did not run LobeHub.

• Agent Builder that creates agents from a one-sentence description
• Agent Groups that run tasks in parallel, plus Schedules, Projects, and a team Workspace
• IM Gateway that connects agents to messaging platforms
• Personal Memory with Continual Learning that users can inspect and edit
• 10,000+ tools and MCP-compatible plugins
• One-click deployment to several clouds, Docker, and database-backed server mode

• One of the most polished open chat interfaces, with a large community (about 83,000 stars)
• Agent groups, schedules, and projects go beyond single-turn chat
• Editable memory gives users control over what the assistant retains
• Many deployment paths, including serverless and Docker

• The LobeHub Community License allows commercial use as a frontend and backend service only without modifying the source, and a derivative work needs a commercial license from the producer
• The license text also lets the producer make the open-source terms stricter or more relaxed, and it does not address branding or user counts
• The README says the plugin system is still under major development
• The official self-hosting documentation was a placeholder when we checked, so setup details live in the README and community guides
• Requires your own model API keys (OPENAI_API_KEY at minimum) and gives no hosted pricing on the pages we read

Self-hosting is free, and you pay for model APIs and any hosting. The hosted app at app.lobehub.com exists, but the pages we reviewed show no prices. Commercial licensing for derivative works is arranged with the producer (hello@lobehub.com). The OpenAlternative page also lists an affiliate VPS offer, which is not LobeHub pricing.

ChatGPT, Claude, and Gemini are hosted assistants with first-party apps and models. LobeHub gives you a similar chat experience across many providers and adds agent groups, scheduling, and editable memory on a stack you can host yourself. Compared with Open WebUI, which focuses on being a self-hosted gateway to local and cloud models and has its own branding condition, LobeHub leans toward consumer-friendly polish and agent features, with a license that restricts modification for commercial use. Cherry Studio and AnythingLLM are desktop-first alternatives.

• People who want a polished multi-model chat interface they can host unmodified
• Teams experimenting with agent groups, schedules, and shared workspaces
• Users who want to inspect and edit what their assistant remembers

• Organizations that need to modify the source for a commercial product without buying a license
• Teams that need a standard OSI-approved license
• Anyone who wants a minimal, stable interface with few moving parts