ByteDance's self-hosted "super agent" for deep research, reports, slides and web pages, with Docker sandboxes, sub-agents, scheduled tasks and chat-app bots.
DeerFlow 2.0 (83.9k stars, MIT, v2.1.0 September 2026) is a web workspace that runs a lead agent with sub-agents, memory and code sandboxes (local, Docker or Kubernetes). Built-in skills cover deep research, report writing, slide decks, web pages, charts, image, video and podcast generation; it also has scheduled tasks, MCP plugins and bots for Telegram, Slack, Discord, Feishu and others. You bring model keys (it suggests Doubao, DeepSeek or Kimi; vLLM works) and a search provider such as Tavily or InfoQuest. Needs at least 4 vCPU and 8 GB RAM; Linux plus Docker is the recommended server.
Cite this pageRepoQuarry. “DeerFlow.” Checked 16 Oct 2026. https://repoquarry.dev/projects/deer-flow
How much of Manus does it replace?
5 of 6 jobs. Covers research, sandboxed work, slides, schedules and connectors when self-hosted; no cloud browser product or hosted website publishing.
No:Autonomous web browsingNot covered
Partial:Deep research reportsPartial replacement
Partial:File and code tasks in a sandboxPartial replacement
Partial:Slides generated from a promptPartial replacement
Partial:Scheduled agent tasksPartial replacement
Partial:Connectors to your appsPartial replacement
For autonomous web browsing, see other Manus alternatives. You may also need to build an interface, an API server and hosting yourself.
How much of Genspark does it replace?
4 of 4 jobs. Research, slides and sandboxed tasks, with Slack, Telegram and other chat bots; you run and configure it yourself.
Partial:Slides generated from a promptPartial replacement
Partial:Deep research reportsPartial replacement
Partial:File and code tasks in a sandboxPartial replacement
Partial:Connectors to your appsPartial replacement
Facts, with sources
Every answer links to where we found it. We read the documentation; we did not install it.
Runs on my machine?Confirmed:Self-hosted with Docker (recommended) or a local dev setupLinux plus Docker is the recommended deployment target for a persistent server.github.com · deer-flow#deployment-sizing ↗today
Model weights licenceConfirmed:No separate model weights; you bring model API keys or host a model yourself (vLLM supported)If you also host a local LLM, size that service separately.github.com · deer-flow#deployment-sizing ↗today
Commercial useConfirmed:Yes, MIT code; model and search provider terms apply separatelygithub.com · LICENSE ↗today
Paid services needed?Partial:Needs a model (API key or self-hosted) and a web search provider such as Tavily or InfoQuest; self-hosted fetchers existTAVILY_API_KEY=your-tavily-api-keygithub.com · deer-flow#configuration ↗today
HardwareConfirmed:Minimum 4 vCPU, 8 GB RAM, 20 GB free SSD for local use (more for Docker or shared servers); a local LLM is sized separately4 vCPU, 8 GB RAM, 20 GB free SSDgithub.com · deer-flow#deployment-sizing ↗today
Operating systemsConfirmed:Linux recommended for servers; macOS and Windows (Git Bash) for development or evaluationmacOS and Windows are best treated as development or evaluation environments.github.com · deer-flow#deployment-sizing ↗today
Setup effortConfirmed:Medium; make setup writes config, then make docker-start; needs Docker Compose v2.24+ and model and search keysRequires Docker Desktop / Docker Engine and Docker Compose v2.24+github.com · deer-flow#option-1-docker-recommended ↗today
Your choice on cookies
We use essential storage to keep the site secure and remember this choice. With your permission we'd also count visits to see which pages help people, and remember your filters. We never use advertising or cross-site tracking cookies, and sponsors never get your data. Cookie policy