Manus
Manus (A general AI agent that takes a goal and does the work for you, from research and websites to slides, videos and games) is a general AI agent: instead of chatting back and forth, you give it a goal and it plans and carries out the whole task on its own, working in its own cloud environment where it can browse the web, write and run code, and build things.
It went viral in early 2025 as one of the first truly general-purpose AI agents, and it can produce slide decks, websites, games, videos, research reports, and data analysis end to end. Manus is made by Butterfly Effect, a company founded in Beijing in 2022 and now headquartered in Singapore, previously known for the Monica browser assistant. Its founders are Xiao Hong (CEO) and Ji Yichao. The company raised at a roughly 500 million US dollar valuation in a 2025 round led by Benchmark, and in late 2026 raised a further 500 million US dollars or more, after a proposed Meta acquisition was blocked.
This page explains what Manus is, how it works, its pricing and credits, honest reviews, and how it compares with Genspark.
Overview
What is Manus?
Manus is an autonomous AI agent, a step beyond a chatbot. Where a normal assistant answers and waits, Manus takes a goal, breaks it into steps, and executes them in its own virtual computer in the cloud: it opens a browser, searches and reads, writes and runs code, creates files, and assembles a finished result, working asynchronously so you can leave it running and come back to the output. It became famous in early 2025 as one of the first products to show a general agent doing real, multi-step work, and it is used to build websites, slide decks, games, videos, data analyses, and research reports from a single instruction.
Manus is built by Butterfly Effect, a startup founded in Beijing in 2022 and now headquartered in Singapore, which was previously known for the Monica AI browser assistant. The company and its agent drew enormous attention, a reported valuation around 500 million US dollars after a 2025 round led by Benchmark, a proposed acquisition by Meta in late 2025 that was ultimately blocked, and a further large funding round in late 2026. Its founders are Xiao Hong, the CEO, and Ji Yichao.
Manus company information
| Detail | Information |
|---|---|
| Product | Manus (general AI agent) |
| Company | Butterfly Effect |
| Founders | Xiao Hong (CEO) and Ji Yichao |
| Founded | 2022 (Beijing); HQ now Singapore |
| Also known for | Monica (AI browser assistant) |
| Funding | ~$500M valuation (2025, Benchmark-led); further $500M+ raised late 2026 |
| Category | Agentic AI / autonomous agent |
| Website | manus.im |
How Manus works
You describe an outcome in plain language, for example "research our top three competitors and build a one-page website comparing us." Manus spins up a cloud workspace, plans the task, and does it: browsing sources, writing and running code, generating files, and showing its steps as it goes. Because it runs asynchronously, long jobs continue in the background and notify you when done, which is different from a chatbot that replies instantly. Work is measured in credits, and the free plan refreshes a daily allowance, while paid plans give monthly credit pools. Complex, long-running tasks consume more credits, so heavier use needs a higher tier.
Key Manus features
Manus is a do-it-for-you agent rather than a single-purpose tool. These are the capabilities people use most.
- Autonomous task execution: plans and completes multi-step jobs end to end in a cloud virtual computer.
- Web browsing and research: searches, reads, and synthesises sources into reports and analyses.
- Builds websites, apps and games: writes and runs code to produce working deliverables.
- Slides, documents and data: creates presentations, documents, and spreadsheets with the content filled in.
- Video and design: generates video and design assets as part of a task.
- Apps and API: available on web, desktop, iOS, and Android, with an API for developers.
Manus pricing
Manus is freemium and credit-based. The Free plan gives a daily refresh of around 300 credits so you can try the agent. Paid plans are reported roughly as a Starter tier near 19 to 20 US dollars a month with about 3,900 to 4,000 credits, a Plus tier near 39 to 40 US dollars a month with about 8,000 credits, and a Pro tier near 199 to 200 US dollars a month with about 40,000 credits, plus Team per-seat plans. There is also a separate Cloud Computer add-on that some sources list from around 30 US dollars a month. The important point for budgeting is that long, complex autonomous tasks consume a lot of credits, so your real capacity depends on your workload. Figures vary across sources and change, so confirm current plans on manus.im before subscribing.
Manus reviews: the honest picture
Manus earned its hype by showing a general agent actually completing complex work, and reviewers credit it with strong performance on multi-step tasks and good scores on agent benchmarks, along with polished outputs like websites and decks. Its iOS app holds a rating around 4.7 stars. The honest caveats are familiar for this category. First, it is credit-hungry: long autonomous runs can burn through credits quickly, so heavy use gets expensive. Second, those long runs can be slow, and the agent occasionally stalls, takes a wrong path, or needs re-running. Third, because it operates autonomously in the cloud, you should review what it produces rather than trust it blindly. The fair summary: Manus is one of the most capable autonomous agents available and genuinely useful for hands-off tasks, with credit cost and occasional reliability hiccups as the trade-offs.
Manus apps and links
| Channel | Link |
|---|---|
| Website | manus.im |
| iOS (App Store) | Manus on the App Store |
| Android (Google Play) | Manus on Google Play |
| Documentation | manus.im/docs |
| X (Twitter) | x.com/manusai |
Manus vs Genspark
Manus and Genspark are the two headline AI super agents of this era, and people often weigh them against each other. Both take a goal and work autonomously to a finished result, and both are credit-based with the same core trade-off of impressive autonomy against credit cost. The difference is emphasis. Manus is positioned as a more open-ended general agent that spins up its own cloud computer to browse, code, and build almost anything, which suits open-ended, build-heavy tasks. Genspark is a broader productivity workspace with strong built-in outputs (slides, sheets, docs, research pages) and consumer features like its Call For Me phone-call agent. If you want an open-ended autonomous builder, Manus is compelling; if you want polished documents and research in one workspace, Genspark fits better. Many power users try both. The table sums it up.
Who should use Manus
Manus suits founders, developers, analysts, and operators who have open-ended, multi-step tasks they would rather hand off entirely, building a site, running research, prototyping an app, than do step by step. It is less suited to users on a tight budget who run many long tasks, since credits add up, or to anyone who needs guaranteed, deterministic output without review. Used for high-value, hands-off jobs where the time saved justifies the credits, it is one of the most capable autonomous agents on the market in 2026.
History
Manus was founded in 2022 in Singapore (founded in Beijing) by Xiao Hong (Co-founder & CEO) and Ji Yichao (Co-founder).
Manus features
Manus offers the following capabilities:
- Autonomous task execution. Plans and completes multi-step jobs end to end in its own cloud virtual computer.
- Web browsing and research. Searches, reads, and synthesises sources into structured reports and analyses.
- Builds websites, apps and games. Writes and runs code to produce working deliverables, not just descriptions.
- Slides, documents and data. Creates presentations, documents, and spreadsheets with the content filled in.
- Runs asynchronously in the cloud. Leave long tasks running in the background and get notified when they finish.
- Apps, desktop and API. Available on web, desktop, iOS, and Android, with an API for developers.
Use cases of Manus
- Hand off multi-step tasks (Founders and operators). Delegate open-ended jobs and get a finished deliverable back without doing each step yourself.
- Build websites and prototypes (Makers and developers). Have the agent write and run code to produce a working site, app, or prototype.
- Run research and analysis (Analysts). Turn a research goal into a sourced report or data analysis with minimal supervision.
- Produce decks and docs (Consultants). Generate presentations and documents end to end, then refine rather than build from scratch.
- Automate repetitive workflows (Teams). Offload recurring, multi-step tasks to an autonomous agent to save hours.
Manus pricing
Manus uses a freemium pricing model, starting at Free; paid from ~$19/mo. The product offers a free version.
| Plan | Price | What's included |
|---|---|---|
| Free | $0 /daily credits | A daily refresh of around 300 credits so you can try the agent on smaller tasks.
|
| Starter | ~$19-20/mo /subscription | About 3,900 to 4,000 credits a month for regular users running moderate tasks.
|
| Plus | ~$39-40/mo /subscription | About 8,000 credits a month for frequent users with heavier or longer tasks.
|
| Pro | ~$199-200/mo /subscription | About 40,000 credits a month for power users and teams running many long autonomous jobs.
|
Manus vs Genspark
The table below contrasts Manus with Genspark (www.genspark.ai) on a few key dimensions.
| Dimension | Manus | Genspark |
|---|---|---|
| Type | Open-ended general agent | Broad agentic workspace |
| How it runs | Own cloud virtual computer | Mixture of models and tools |
| Built-in outputs | Sites, apps, decks, research | Slides, sheets, docs, research |
| Standout | Open-ended build + browse + code | Call For Me (real phone calls) |
| Apps | Web, desktop, iOS, Android | Web, iOS, Android |
| Best for | Open-ended autonomous building | Finished docs, decks, research |
Platforms & tech
- Supported platforms: Web browser, iOS, Android, Desktop.
- Deployment: ios, android, cloud.
Awards & recognition
- One of the first viral general AI agents — Launched in early 2025 by Butterfly Effect, Manus became one of the first widely adopted general AI agents. It reached a reported ~$500M valuation in a Benchmark-led round, drew a proposed Meta acquisition that was blocked, and raised a further $500M+ in late 2026.
See also
- Genspark — The other headline AI super agent, a broad workspace with strong built-in slides, sheets, and research plus Call For Me.
- ChatGPT — OpenAI's assistant with its own agent mode and tools, the most general-purpose alternative.
- Perplexity — An AI answer engine strong at sourced research, lighter-weight than a full autonomous agent.
References
- "Manus official website". manus.im.
- "Manus (AI agent)". Wikipedia.
- "Manus on the App Store". Apple App Store.
- "Manus AI review and pricing". Taskade.
External links
Frequently asked questions
- What is Manus?
- Manus is a general AI agent: you give it a goal and it plans and completes the whole task autonomously in its own cloud environment, where it can browse the web, write and run code, and build things. It can produce websites, slide decks, games, videos, research reports, and data analyses end to end. It is made by Butterfly Effect, a company founded in Beijing in 2022 and now based in Singapore, and it became famous in early 2025 as one of the first general-purpose AI agents.
- How much does Manus cost?
- Manus is freemium and credit-based. The Free plan gives a daily refresh of around 300 credits. Paid plans are reported roughly as Starter near 19 to 20 US dollars a month (about 4,000 credits), Plus near 39 to 40 US dollars a month (about 8,000 credits), and Pro near 199 to 200 US dollars a month (about 40,000 credits), plus Team per-seat plans and a separate Cloud Computer add-on. Long autonomous tasks consume many credits, so check manus.im for current plans and match the tier to your workload.
- Is Manus good? What do reviews say?
- Manus is widely regarded as one of the most capable autonomous agents, praised for completing complex multi-step tasks and for polished outputs like websites and decks, with strong scores on agent benchmarks and an iOS rating around 4.7 stars. The main criticisms are that it is credit-hungry so heavy use gets expensive, that long runs can be slow and occasionally stall or take a wrong path, and that autonomous output should be reviewed. It is powerful and genuinely useful, with cost and occasional reliability as the trade-offs.
- Who owns Manus?
- Manus is made by Butterfly Effect, a startup founded in Beijing in 2022 and now headquartered in Singapore, previously known for the Monica AI browser assistant. Its founders are Xiao Hong, the CEO, and Ji Yichao. The company reached a reported valuation around 500 million US dollars in a 2025 round led by Benchmark, was the subject of a proposed Meta acquisition in late 2025 that was ultimately blocked, and raised a further 500 million US dollars or more in late 2026.
- Manus vs Genspark: which is better?
- Both are leading AI super agents that work autonomously and are credit-based. Manus is positioned as a more open-ended general agent that spins up its own cloud computer to browse, code, and build almost anything, which suits open-ended, build-heavy tasks. Genspark is a broader productivity workspace with strong built-in outputs (slides, sheets, docs, research) and consumer features like its Call For Me phone-call agent. Choose Manus for open-ended autonomous building and Genspark for finished documents and research in one workspace. Many power users try both.
- How is Manus different from ChatGPT?
- ChatGPT is primarily a conversational assistant that answers and can use tools, while Manus is built from the ground up as an autonomous agent that executes whole tasks on its own in a cloud environment, browsing, coding, and producing finished deliverables with less step-by-step guidance. ChatGPT now has its own agent mode, so the lines blur, but Manus's focus is hands-off, end-to-end task completion rather than chat. Which is better depends on whether you want a conversational assistant or an autonomous task agent.
- Does Manus have a mobile app?
- Yes. Manus is available as the "Manus - AI Agent & Automation" app on both the iOS App Store and Google Play, in addition to the web and a desktop app. The iOS app holds a rating around 4.7 stars from tens of thousands of ratings. The apps let you launch and monitor autonomous tasks on the go, with results synced to your account across devices, which fits its asynchronous, run-in-the-background model well.
- Why do Manus credits run out quickly?
- Because Manus actually performs long, multi-step tasks rather than just replying, each run, browsing, coding, building, generating files, consumes credits, and a single complex job can use a large number. That is why heavy users report credits depleting fast. The practical advice is to size your plan to your workload, reserve long autonomous runs for high-value tasks, and watch your remaining balance so a few big jobs do not exhaust your monthly or daily allowance.
- What are the best Manus alternatives?
- The closest alternative is Genspark, the other prominent AI super agent, which is a broad workspace with strong built-in slides, sheets, and research plus a Call For Me phone agent. For a general assistant with an agent mode, ChatGPT is the most versatile, and for sourced research with less autonomy, Perplexity is lighter. The right choice depends on whether you want open-ended autonomous building (Manus), finished productivity outputs (Genspark), a general assistant (ChatGPT), or fast research (Perplexity).