Repository navigation
manus alternatives
People search for a Manus alternative for two different reasons, and they need different answers. The first group hit a limit in the product: cost, task fit, or the ceiling any hosted agent has. The second group watched the ownership news - acquired by Meta for about $2 billion in December 2025, ordered unwound by Chinese regulators in April 2026, bought back by a Tencent-led consortium by August - and decided they would rather not build a workflow on top of a company whose owner changed three times in a year. Both groups end up at the same fork: another hosted general agent, or open software you run yourself. This page walks both prongs, plus the section most Manus comparisons skip: the open-source component that was found doing Manus's browsing in the first place.
The disclosure that this wiki attaches to every comparison: invisible_playwright_mcp, whose wiki this is, appears in the open-source list below. Facts about Manus and every other tool were checked against their own pages or press coverage on 2026-09-03, and the sources are at the bottom.
The product worked. Manus launched on 6 March 2025, went viral off an agent demo, and by its own blog post of 17 December 2025 had crossed $100M in annual recurring revenue eight months after launch, with total run-rate above $125M and growth of more than 20% month over month. The same post counts more than 80 million virtual computers created for tasks. Whatever else is true, the demand it found was real.
The ownership did not hold. Butterfly Effect, Manus's Chinese-founded, Singapore-incorporated parent, agreed to a roughly $2 billion acquisition by Meta in December 2025. Around April 2026, Chinese regulators ordered the deal unwound on national security grounds, citing technology export controls and foreign investment rules. By June, Meta had cut Manus off from its internal systems and halted data sharing; by July, Tencent was in talks to become the largest outside shareholder; in August, a Tencent-led consortium including HongShan bought the company back at the original price and Manus returned to operating as an independent company.
Why it matters for your choice. None of that stopped the product from shipping. But if you are choosing an agent to embed in real workflows, the episode is a live demonstration of the thing this wiki keeps writing about hosted agents generally: the product you rent sits inside somebody's corporate strategy, and regulators, acquirers and boards can all move it without asking you. With Manus the data question was explicit - part of the unwinding was Meta halting data sharing between the companies - and the question "who exactly holds my sessions and files now?" got three different answers in nine months.
Manus is a hosted general agent: you give it a goal, it plans, spins up a cloud virtual computer, browses, writes files, runs code, and comes back with results. That is a wider brief than a pure browser agent, and it is the frame for comparing anything to it. An alternative needs to answer at least the browsing-and-acting half; the closer it also gets to files, code and long multi-step plans, the closer it is to the whole product.
This deserves its own section because the two names are entangled in a way most comparison pages get wrong. In March 2025, shortly after the viral launch, an X user named Jian got Manus to reveal its sandbox runtime and found it running Claude Sonnet with access to 29 tools, using the open-source library browser-use for its web browsing. Manus's chief researcher Yichao "Peak" Ji confirmed the spirit of the finding, saying the agent "relies heavily on open-source technologies and wouldn't exist without open source".
So "browser use vs Manus" is not two rival products; it is a component versus a product built above that layer. browser-use is an MIT-licensed Python library, about 112k GitHub stars as of this writing, that connects an LLM of your choice to a Chromium-family browser. Manus at the time wrapped that layer, and much else, in a planner, a cloud sandbox and a subscription. If what you valued in Manus was the browsing, the raw ingredient is free and self-hostable, and you assemble the rest. If what you valued was not having to assemble anything, a library is not an alternative for you, and the hosted rows below are. browser-use has since grown its own cloud offering as well, which sits somewhere between the two - browser-use alternatives covers its trade-offs properly.
ChatGPT Work (OpenAI, launched 9 July 2026) is the closest current big-vendor analogue to Manus's brief: an agent that stays with a multi-step task across apps and files and returns finished documents and reports. It is deliverable-shaped rather than watch-it-click-through-a-site shaped, and it lives entirely inside OpenAI's product strategy, whose recent churn is documented on Is OpenAI Operator still available?.
Gemini Agent (Google, in the Gemini app since 18 November 2025) does multi-step tasks that mix live web browsing with Google's tools, launched for AI Ultra subscribers in the US and spreading down tiers during 2026. It inherits Project Mariner's browser control work; that lineage and what else Google offers is on Project Mariner is gone: what replaced it.
Every row here shares the property the Manus story illustrated: your sessions, files and history live with the vendor, and continuity depends on decisions you do not get a vote in.
Stars and licenses below were read from each repository on 2026-09-03.
| Project | Stars | License | What it is |
|---|---|---|---|
| OpenManus | ~58k | MIT | A community general agent built in Manus's image after the launch, when Manus was invite-only |
| browser-use | ~112k | MIT | The browsing layer itself; the component found inside Manus |
| Skyvern | ~23k | AGPL-3.0 | Vision-first browser workflows on Playwright, with a managed cloud |
| Agent S | ~12k | Apache-2.0 | Whole-desktop computer use, the widest environment of the open set |
| invisible_playwright_mcp | ~30k | MIT | Ours: an agent bound to its own source-patched Firefox |
OpenManus is the most literal alternative in the list: it exists because Manus's launch was invite-only and the community built an open one. The others each pick a different piece of the brief. What none of them gives you is Manus's managed cloud sandbox; what all of them give you is code you can read, run where you like, and keep through anybody's acquisition.
Where invisible_playwright_mcp fits, stated by the people who make it: its bet is the browser rather than the planner. It drives a Firefox patched at the source level so that what a page inspects reads as a normal desktop machine, it runs on your hardware with your OpenRouter key or inside an MCP assistant you already use, and its scope is browsing tasks, not Manus's files-and-code sandbox. Windows and Linux only, and no promise of guaranteed passage anywhere - that boundary is documented, not waved away.
- You want the full plan-browse-code-files loop, managed for you. Manus remains the incumbent at that brief; ChatGPT Work is the big-vendor rival. Read the ownership paragraph again before you embed either deeply.
- You mostly need the browsing half. An open browser agent covers it on your own machine; start from the table and Choosing an AI browser agent.
- You want Manus's shape but open. OpenManus, plus your own model key and patience: community recreations trail the product they mirror.
- Your tasks touch accounts and data you cannot send to a third party. That rules out the hosted rows entirely, and it is the strongest single reason this page's open half exists.
Is Manus still available? Yes. It operates as an independent company again after the Meta acquisition was unwound; a Tencent-led consortium bought it back at the original price in August 2026.
Did Meta buy Manus? It announced the roughly $2B acquisition in December 2025 and completed it, then unwound it after Chinese regulators ordered divestiture in April 2026 on national security grounds.
Is Manus built on browser-use? In March 2025 its sandbox was found running the open-source browser-use library among 29 tools, and its chief researcher said the agent "wouldn't exist without open source". The company has shipped substantially since, so treat that as its documented starting point rather than a statement about today's internals.
What is the best free Manus alternative? OpenManus is the closest in shape; browser-use is the strongest single component; both are free software, and you pay only for model tokens.
Is there a Manus alternative that keeps data on my machine? Any of the open-source rows: they run locally, and the model you point them at is the only outside party you introduce.
Is Manus free, or is there a free tier? It is a paid product with a limited free plan below it, and the paid tiers are where the useful capacity sits. Prices in this category move often enough that a number written here would be wrong within a quarter, so check theirs; what is actually free in the agent stack explains why "free" means four different things across this field and which one usually applies.
Claude Code versus Manus? Not the same shape of thing. Claude Code is a coding assistant that runs on your machine and can be given tools, including a browser through MCP; Manus is a hosted autonomous agent you send a task to. If what you want is an agent that works inside a tool you already run, that is the MCP route rather than either of them: the MCP server.
How do I cancel Manus? Through Manus's own account settings, and their support is the only accurate source for it - we do not run the product and will not guess at somebody else's billing flow. If you are leaving because you want the agent local and keyed to your own provider, the open-source route above is the part of this page worth reading.
Is invisible_playwright_mcp a Manus replacement? Only for the browsing part of the brief, and we maintain it, so get a second opinion from the table above.
See also: browser-use alternatives for the component layer, Project Mariner is gone: what replaced it for the parallel story at Google, and What is an AI web agent? if the category itself is new to you.
- Manus: $100M ARR, $125M revenue run-rate, fetched 2026-09-03: the revenue, growth and virtual-computer figures, dated 17 December 2025.
- TechCrunch: Meta reportedly moves to unwind $2B Manus deal after Beijing's demand, fetched 2026-09-03: the December 2025 deal, the April 2026 divestiture order and grounds, the systems cutoff and data-sharing halt, the Singapore relocation.
- Crypto Briefing: Tencent leads effort to unwind Meta's $2B Manus acquisition, fetched 2026-09-03: the Tencent and HongShan consortium and the buyback at the original price. CNBC: Manus to return as independent company and Bloomberg: Tencent in talks to take big Manus stake, surfaced via search 2026-09-03.
- The Decoder: Chinese AI agent Manus uses Claude Sonnet and open-source technology, fetched 2026-09-03: the March 2025 finding and the chief researcher's quotes.
- Wikipedia: Manus (AI agent), surfaced via search 2026-09-03, for the 6 March 2025 release date.
- The OpenManus, browser-use, Skyvern, Agent-S and invisible_playwright_mcp repositories, stars and licenses read via the GitHub API 2026-09-03.
- 9to5Google on Gemini Agent's launch, fetched 2026-09-03; ChatGPT Work launch date per Bloomberg, surfaced via search 2026-09-03.
Written by the maintainers of invisible_playwright_mcp, one of the open-source rows above. The Manus ownership saga is genuinely useful to our argument, which is precisely why every beat of it is cited to the reporting rather than told from memory.
- OpenAI Operator alternatives
- Open-source Operator-style agents
- Is OpenAI Operator still available?
- OpenAI Operator vs Claude computer use
- browser-use alternatives
- Choosing an AI browser agent
- Open-source AI browser agents
- Open-source computer-use agents
- What is an AI web agent?
- AI browser agents vs traditional scraping
- Cloud browser infrastructure for AI agents, explained
- Browserbase alternatives
- Firecrawl vs an AI browser agent
- Skyvern alternatives
- Stagehand vs browser-use
- Project Mariner is gone: what replaced it
- Manus alternatives
- Gemini computer use vs Claude computer use
- A stealth browser MCP, reviewed honestly by its own wiki
- AI browser vs AI browser agent: which one do you want?
- AI browser agent vs RPA: which one fits the job
- AI browser agent vs n8n, Zapier and Make
- Vercel agent-browser alternatives, compared honestly
- What is an agentic browser? Definition and the two kinds
- Open-source agentic browsers: the three layers, compared
- Choosing an MCP server for browser automation: four axes
- Stealth MCP servers compared: Camoufox, nodriver, Patchright
- Playwright MCP alternatives, and the three you don't need
- Autonomous browser agents: the four rungs of autonomy
- What is actually free in the AI browser agent stack
- browser-use on GitHub: what the repo actually gives you
- Playwright MCP vs Chrome DevTools MCP: different jobs
- How to choose among MCP servers: a map by category
- Which MCP servers are worth adding to Claude Code
- MCP on GitHub: finding servers and judging them fast
- MCP vs an API: the decision, and what the wrapper costs
- MCP alternatives: when the protocol is the wrong shape
- Why does my AI agent get blocked?
- The timing signal AI agents give off
- Agent retry loops trip rate limits, not fingerprints
- Claude computer use detected as a bot
- browser-use getting blocked: what you can and cannot change
- Playwright MCP session blocked: four causes, four fixes
- Playwright MCP and captchas: what actually gets you past
- Cloudflare and a browser MCP server: what is being read
- Can an AI agent solve a captcha? The honest answer
- Getting an AI agent to fill out forms
- The best model for an MCP browser agent, and what it really costs
- Browser problem or model problem?
- How to use a Playwright MCP server with Claude Code
- Extracting data to a CSV with an AI agent
- Monitoring a page for changes with an AI agent
- How to let Claude Desktop control a browser
- How to add a browser to Cursor as an MCP server
- Using an AI agent to hunt for apartments
- Getting website data into Google Sheets with an AI agent
- Using an AI agent to download invoices from portals
- AI agents for web research
- Using an AI agent to test your own website
- How to add a browser to Cline as an MCP server
- Posting to social media with an AI agent
- Posting to Facebook with an AI agent
- Posting to Instagram with an AI agent
- Posting to X with an AI agent
- Automating LinkedIn posts: read this first
- Appointment bots: what they are and what an agent can legitimately do
- Track prices across sites with an AI agent
- Build a lead list with an AI browser agent
- Run an AI browser agent on a schedule
- AI browser agent with a local LLM: what changes
- Should you log your AI agent into your accounts?
- How to write a task an AI browser agent can follow
- Move data between two web apps with an AI agent
- The MCP server
- How the tools are shaped, and why
- Playwright MCP vs the Playwright CLI: which fits when
- Playwright MCP: browser is already in use, and the fix
- Playwright MCP best practices: four decisions that matter
- Playwright MCP with a proxy, and the three leaks it leaves
- A browser MCP server in GitHub Copilot: setup and limits
- Using a browser MCP server for web scraping: the pattern
- Which LLM for browser automation: the four properties
- How to build a browser agent, and what to take instead
- Getting an AI agent to log into a website: three routes
- MCP tools, resources and prompts: who controls each
- How many MCP tools is too many? The context arithmetic
- How to build an MCP server: the decisions, not the scaffold
- Local or remote MCP server: what changes, and what does not
- Writing an MCP client in Python: the thirty-line version
- Self-hosted AI agent: what one actually costs to run
- How long an AI browser agent takes per step, measured
- Text, HTML, snapshot or screenshot: what the agent should read
- Giving an AI browser agent a stopping condition
- Keeping an AI browser agent out of destructive actions
- Why did the AI agent click the wrong thing
- When the page changes under the AI agent
- Running one AI agent task across a list of sites
- Seeing a page as it appears in another country
- Getting data out of a dashboard with no export button
- Two browsers in one session: main and support
- Finding the dead links on a site with an AI agent
- Filling a CRM record from a company's website
- One form submission per spreadsheet row, with an AI agent
- Dated screenshots of a page as evidence
- Checking order and delivery status with an AI agent
- Reading a PDF that opens inside the browser
- Summarising a long page or thread with an AI agent
- Collecting every image on a page with its caption
- Collecting event and course listings with an AI agent
- Cancelling a subscription with an AI agent
- What an AI agent can and cannot do inside an iframe
- Shadow DOM and an AI agent: you can click it, you cannot read it
- How to add a browser to Codex as an MCP server
- What a page snapshot costs, per control
- How to let Gemini CLI use a browser
- Native selects and the ones that only look like selects
- Clicking by selector or by coordinates
- How long the agent waits before it gives up
- What a second browser costs
- Uploading a file with an AI agent, and why this one cannot
- Watching the agent work, and when it is worth it
- When not to use an AI browser agent
- Agent or script: deciding once instead of every time
- Using the keyboard instead of the mouse
- Secrets in an agent task: where they end up
- What an agent run should log
- Deduplicating what an AI agent collects
- Normalising values across sites
- Validating an AI agent's output
- Reading a table with an AI agent
- Driving a site's own search and filters
- The task works headed and fails headless
- How to let the AdaL CLI use a browser