LobeHub is an AI agent platform for creating agents, grouping them into teams and scheduling their tasks on web, desktop and mobile, available as a credit-based cloud service or a self-hosted community edition.
Used this tool? Rate it
Used this tool? Rate it
LobeHub is an AI agent platform for building agents, grouping them into teams and running their tasks on web, desktop and mobile. The company calls it a collaborative agent platform for work and life. Its homepage pitch is that LobeHub organizes your agents into 7×24 operation and hires, schedules and reports on your entire AI team.
You can use it in two ways: as LobeHub Cloud, a managed service billed in credits, or as a self-hosted community edition that you deploy on your own servers. The service and its Terms of Service are run by LobeHub LLC.
LobeHub grew out of LobeChat, an open-source multi-model chat interface. The company says its vision began with LobeChat's multi-model support. The repository's license file still uses the old name: from 1.0, LobeChat is licensed under the LobeHub Community License, based on Apache License 2.0 with additional conditions.
Version 2.0 marked the turn away from a chat client, and its release notes describe LobeHub v2.0 as a practical shift from single-agent chat to real multi-agent collaboration. Older editorial coverage describes Lobe Chat as a local and privacy-focused ChatGPT-like UI framework, which differs from the cloud-first v2.x architecture described under Limitations.
LobeHub treats agents as the unit of work: you create agents, combine them into groups, give them memory and tools, and hand them tasks that can run on a schedule.
The two approaches differ in who assembles the team. In a group you pick the members and the collaboration mode yourself; CAO decides on its own when to bring in sub-agents and keeps the main plan in its hands.
Two documented workflows are specific enough to follow step by step: setting up a recurring agent task and moving data out of the old desktop local database.
If you already upgraded, the v1.x local database files remain on disk until you remove them, so you can reinstall a v1.x build and export afterwards. Agents, conversations and settings can all move over, typically in a few minutes.
LobeHub's own showcase and documentation examples center on research, writing and coordination work that benefits from several agents or repeated runs.
LobeHub targets individuals and teams that want agents to carry ongoing work, with separate routes for self-hosters and organizations that need private deployment.
It is a weaker fit in these cases:
You can use LobeHub on the web, desktop and mobile, signing in with the same account across them.
LobeHub can be self-hosted on your own infrastructure, giving you control over your data, customization and deployment environment. A self-hosted deployment needs at least one AI provider API key. Server deployments require PostgreSQL 14 or later. Documented deployment methods include Vercel, Docker, Docker Compose, Zeabur, Sealos, RepoCloud and Dokploy.
LobeHub Cloud is billed in credits, and LobeHub converts USD 1 of model usage into 1,000,000 credits for calculation purposes.
| Plan | Monthly billing | Billed yearly (per month) | Credits per month | File storage |
|---|---|---|---|---|
| Free | 0 USD | 0 USD | 500,000 | 15 MB |
| Starter | 12.9 USD | 9.9 USD | 5,000,000 | 1024 MB |
| Premium | 24.9 USD | 19.9 USD | 15,000,000 | 2048 MB |
| Ultimate | 49.9 USD | 39.9 USD | 35,000,000 | 4096 MB |
On the pricing page itself, Starter appears as $12.9 per month or $118.8 per year. The page advertises a free trial of GPT-6.1 Sol, Claude Sonnet 5.5 and MiMo-V2.6 Pro with no credit card required. The Free plan card estimates about 70 messages with GLM-5.3-Flash. On the same card, GPT-6 Astra is marked Subscription required. For the Enterprise Edition, buyers contact sales via the pricing page.
The Terms of Service that contain these refund rules were last updated on 2024-03-01.
The closest alternatives depend on which side of LobeHub you need: the self-hosted AI chat interface LobeChat was known for, or the hosted multi-agent operator it has become.
LobeHub is the current name and direction of the project that began as LobeChat. The license file still names the software LobeChat, but version 2.0 turned the chat interface into a multi-agent platform.
The community edition's code is public and can be self-hosted, but it ships under the LobeHub Community License rather than plain Apache 2.0. Commercial use without modifying the source is allowed, while distributing a derivative work needs a commercial license.
The Free plan costs 0 USD and includes 500,000 credits per month and 15 MB of file storage. Some models, such as GPT-6 Astra, require a paid subscription.
Subscription points are reset when each billing cycle ends. Top-up credits are separate, carry a six-month validity on the purchase page and are spent after subscription credits.
The privacy policy says chat content from the mobile app's AI features is not used for model training, and that integrated AI services are reached through enterprise API endpoints that do not train on customer data.
Not with the v2.x desktop app alone, because the local database was removed and cloud models and sync need a network connection. Running local models or self-hosting the community edition are the documented ways to keep model calls or data under your control.
