Lobehub
🤯 LobeHub is your Chief Agent Operator, organizing your agents into 7×24 operations by hiring, scheduling, and reporting on your entire AI team.
Last verified:
What is Lobehub?
LobeHub is an open-source AI platform and next-generation agent harness that enables users to create, collaborate with, and evolve long-term agent teammates. Unlike disposable task-driven AI tools, LobeHub builds agents with persistent memory that grow alongside users over time. The platform serves as a comprehensive workspace for work and life, allowing anyone to find, build, and collaborate with agent teams to deliver complex, end-to-end work.
Key features include the Agent Builder for effortlessly creating custom agents with real skill control, Agent Groups for parallel collaboration where multiple agents work together like real teammates, and multi-model orchestration supporting 70+ providers including GPT-4o, Claude, Gemini, DeepSeek, Llama, and Qwen. The platform offers 10,000+ Skills via MCP (Model Context Protocol) to connect agents to tools and APIs, a Resource Library for uploading documents and building knowledge bases, and white-box editable memory for transparent agent learning. Users can create Pages for collaborative writing, Schedule runs for timed agent execution, organize work by Project, and share Workspaces for team collaboration.
LobeHub is designed for developers building custom agents, professionals who use AI agents daily, teams needing flexible AI workspaces, and everyday users who want to craft their own agent teammates. With cross-platform availability (web app, desktop app for macOS/Windows, and mobile apps for iOS/Android), the platform supports voice conversations, image/video generation using DALL-E 3 and Sora technologies, and seamless sync across all devices. The community marketplace features 4,920+ ready-to-use agents, 39,000+ MCP servers, and 210,569+ skills that users can discover, reuse, and remix.
The platform evolved from LobeChat, an open-source chatbot framework, and now has 700,000+ GitHub stars with contributors worldwide. It democratizes token consumption by empowering non-engineers to leverage AI power, offers self-hosting options for the Community Edition, and includes research initiatives like MCPMark for stress-testing MCP benchmarks.
Lobehub pricing
Pricing model: Freemium
LobeHub offers a free tier with 450,000-500,000 compute credits for new users (no credit card required). Paid yearly plans include: Starter at $9.9/month ($118.8/year, 23% off) with 5,000,000 monthly credits, 1GB file storage, 5,000 vector entries, and email/forum support; Premium at $19.9/month ($238.8/year, 20% off) with 15,000,000 monthly credits, 2GB file storage, 10,000 vector entries, and priority email support; Ultimate at $39.9/month ($478.8/year, 20% off) with 35,000,000 monthly credits, 4GB file storage, 20,000 vector entries, and priority chat/email support. All plans include unlimited message requests (bring your own API keys), unlimited chat history, global cloud sync, web search, premium skills, and agent market highlights. Additional credit packages are available for purchase. Enterprise Edition offers commercial license, brand theming, user management, self-hosted providers, private models, and custom integration with custom pricing.
Lobehub pros
- Open-source platform with self-hosting option for Community Edition
- Multi-model support for 70+ AI providers including GPT-4o, Claude, Gemini, DeepSeek
- Persistent white-box memory that is structured and editable by users
- Agent Groups enable parallel collaboration instead of sequential one-by-one execution
- 10,000+ Skills via MCP to connect agents to tools and APIs in one click
- Free tier with 450,000-500,000 compute credits available without credit card
- Cross-platform availability: web, desktop (macOS/Windows), and mobile (iOS/Android)
- Resource Library allows uploading documents for grounded, data-specific answers
- Auto team formation system assembles the right agents for each task automatically
- Schedule runs let agents work at timed intervals even when user is away
- Image and video generation using DALL-E 3, fal.ai, and Sora technologies in-chat
- Voice conversation support with customizable agent voice tones
- Shared Workspaces with clear ownership and visibility for team collaboration
- Community marketplace with 4,920+ agents and 39,000+ MCP servers to discover and remix
- Unlimited chat history and global cloud sync across all devices
- Keyboard shortcuts and command menu for power users to navigate without mouse
- Bring your own provider API keys for unlimited message requests
Lobehub cons
- Paid plans use credit-based system that may be confusing for new users
- Higher-tier models like GPT-4o and Claude 3.7 Sonnet consume credits rapidly
- Free tier credits (450,000-500,000) are limited and reset monthly
- Vector storage is limited even on Ultimate plan (20,000 entries ≈ 200MB)
- File storage caps at 4GB on Ultimate plan, may be insufficient for large teams
- Enterprise edition requires custom pricing and private deployment setup
- Some advanced features like Premium Skills only available on paid plans
- Self-hosting requires technical knowledge for developers to set up properly
Frequently asked questions about Lobehub
Can LobeHub be used for free?
Yes, LobeHub can be used for free in two ways. For professional developers, you can self-deploy the Community Edition which is fully open-source with all capabilities. For the LobeHub Cloud version, all registered users receive 450,000 free computing points ready to use without complex configurations or credit card. If you need more usage beyond the free credits, you can subscribe to Starter, Premium, or Ultimate plans.
What is Compute Unit (Credits)?
Compute Unit (Credits) is a metric used by LobeHub Cloud to measure AI model usage when called upon. Credits are mapped to tokens, and consumption varies for different AI models. For example, GPT-4o mini costs 0.15M credits per 1M input tokens and 0.6M credits per 1M output tokens, while Claude 3.7 Sonnet costs 3M credits per 1M input tokens and 15M credits per 1M output tokens. Your monthly credit allowance determines how many messages you can send with each model.
Will credits be reset monthly?
Yes, compute credits reset monthly. Each plan provides a specific monthly credit allowance: Starter gets 5,000,000/month, Premium gets 15,000,000/month, and Ultimate gets 35,000,000/month. Unused credits do not carry over to the next month. Additional credit packages are available for purchase if you run out before your monthly reset.
How is vector storage calculated?
Vector storage is measured in entries, with approximate file size equivalents. Starter plan includes 5,000 entries (approximately 50MB), Premium includes 10,000 entries (approximately 100MB), and Ultimate includes 20,000 entries (approximately 200MB). Vector storage is used for your knowledge base and resource library, enabling agents to reference your uploaded documents during conversations.
What AI models does LobeHub support?
LobeHub supports 70+ AI providers with 100+ models including GPT-4.1, GPT-4o, GPT-4o mini, ChatGPT-4o, GPT-4 Turbo, OpenAI o1-mini, Claude 3.7 Sonnet, Claude 3.5 Sonnet (New and 0620), Claude 3.5 Haiku, Claude 3 Opus, Claude 3 Haiku, Gemini 2.5 Pro Preview, Gemini 2.0 Flash, DeepSeek V3, DeepSeek R1, Llama, Qwen, and many more. You can switch models mid-conversation or assign different default models to different agents.
Can I bring my own API keys?
Yes, all paid plans support bringing your own provider API keys with unlimited message requests. This means you can use your own OpenAI, Anthropic, Google, or other provider API keys instead of using LobeHub's credit-based system. This is particularly useful for heavy users who already have enterprise API contracts or want more predictable pricing.
What is the difference between Agent Groups and regular agents?
Agent Groups allow you to work with multiple agents like real teammates. The system assembles the right agents for the task automatically, enables parallel collaboration where agents work simultaneously instead of one-by-one, supports multi-task execution without manual coordination, and provides iterative improvement where agents refine outputs together until results are ready. Regular agents work individually, while Agent Groups enable team-based problem solving.
How do I get started with LobeHub?
You can get started in three ways: 1) Use the Web App at app.lobehub.com to start free with no setup needed, 2) Download LobeHub Desktop for macOS or Windows for a native experience, or 3) Download mobile apps from the App Store for iOS or Google Play for Android. Start your first conversation with Lobe AI using the quickstart guide, then create your first custom Agent tailored to your workflow by describing what you need.
What is the Resource Library?
The Resource Library gives your agents real knowledge by allowing you to upload documents, images, spreadsheets, audio, or video. Your agent references these during conversations, so answers are grounded in your actual data rather than just training-set knowledge. You can build a knowledge base for each agent or share resources across your workspace. This is ideal for customer support agents that know your help center, legal assistants trained on your contract templates, or product managers who have read every PRD you've written.
What support channels are available for each plan?
Support channels vary by plan: Starter plan includes Email & Community Forum support; Premium plan includes Priority Email Support; Ultimate plan includes Priority Chat and Email Support. Enterprise Edition includes Custom Integration & support. All users can also access the Docs for answers, join the Discord community, or contact support at [email protected] for technical issues and [email protected] for general inquiries.