Manus
Manus, a Chinese-founded general agent that runs tasks in a visible cloud computer, launches invite-only and goes viral.
Orchestrates models (reportedly Claude 3.5 Sonnet and fine-tuned Qwen) with browser and code tools, claiming to beat OpenAI Deep Research on GAIA. MIT Tech Review found it good on scoped research but slow and crash-prone, at about $2 per task.
- Date
- Thursday, 6 March 2025
- Lab
- Butterfly Effect (Manus)
- Kind
- product
- Access
- research preview
Figures
| Measure | Value | Measured by |
|---|---|---|
| Cost per task (MIT Tech Review test) | ~$2 about one-tenth of ChatGPT Deep Research's price at the time | independent |
Date 2025-03-06 from Wikipedia; the press pieces say 'March 2025'. GAIA claims were Manus's own; invite codes resold for thousands of dollars.
Sources
- www.technologyreview.com/2025/03/11/1113133/manus-ai-review/
- techcrunch.com/2025/03/09/manus-probably-isnt-chinas-second-deepseek-moment/
- en.wikipedia.org/wiki/Manus_(AI_agent)
This record was checked against its sources on 6 October 2026. How we check