Back to vendors
M

Manus

Also known as: Manus, Manus AI, Butterfly Effect, Monica

Visit site
Browser / computer-use agentindependentVerified 2026-07-08

Autonomous general AI agent from Butterfly Effect that runs inside a cloud based virtual computer, taking a high level goal and independently planning and executing multi step tasks, browsing real websites, writing and running code, analyzing data, and returning finished deliverables like reports, spreadsheets, slides, and working web apps.

Manus is an autonomous, general purpose AI agent built by Butterfly Effect, the Singapore based company behind the Monica assistant, that launched in March 2025 and went viral within days for doing something chatbots do not: you give it an objective in plain language and it independently plans the steps, carries them out inside a cloud environment, and returns a finished result. It runs in a Linux style virtual computer where it can use real web browsers, terminals, and file systems, taking screenshots and using vision to verify that its actions completed, which lets it run a longer autonomous horizon than many operator style tools before getting stuck.

The product has matured quickly across versions. Manus 1.5 added full web application generation, producing prototypes with backend logic, a database, and authentication from a single prompt, and Manus 1.6 introduced a faster Chat Mode, a Wide Research feature that runs many sub agents in parallel across sources for deep research, tiered model access through 1.6 Lite, 1.6, and 1.6 Max, plus Scheduled Tasks and connector integrations. It is strongest at open ended, multi step web research and data gathering and at building first draft prototypes, slides, and spreadsheets, and weaker at production grade code and long branching workflows. Ownership drew global attention when Meta announced a roughly two billion dollar acquisition in December 2025, but Chinese regulators blocked the deal in April 2026 and Meta unwound it by June 2026, so Manus continues to operate as an independent standalone service from Singapore, backed by earlier rounds from ZhenFund, HSG, Tencent, and a Benchmark led Series B.

Manus fits researchers, solo builders, and teams that want a hands off agent to run open ended tasks end to end, and is a weaker fit for buyers who need production reliability, predictable cost, self hosting, or who have data governance concerns about a China founded provider operating inside authenticated sessions.

Vendor details

Canonical URL

https://manus.im

Category

Browser / computer-use agent

Subcategory

Autonomous general agent in a cloud virtual computer

Funding status

Independent and based in Singapore, built by Butterfly Effect, the company behind the Monica assistant, founded by Xiao Hong with Ji Yichao as co founder of the Manus product. Earlier funding included a 2023 seed led by ZhenFund, a Series A in November 2024 with HSG, formerly Sequoia China, and Tencent, and a Benchmark led Series B in April 2025 reported at seventy five million dollars at a five hundred million dollar valuation. Meta announced a roughly two to three billion dollar acquisition in December 2025, but China's National Development and Reform Commission blocked it in April 2026 and Meta cut ties by June 2026, leaving Manus independent.

Company status

independent

Use cases & customers

Primary use cases

Autonomous multi step web research and data gatheringBuilding first draft web apps, slides, and spreadsheetsRunning many sub agents in parallel for deep researchScheduled, background task execution in a cloud computer

Target customers

Researchers and analystsSolo builders and solopreneursTeams wanting hands off task automationProduct teams prototyping quickly

Deployment options

Cloud

Integrations

Operates inside a cloud virtual computer with real browsers, terminals, and file systems, so it can browse the live web, run code, and manipulate files directly rather than through narrow plugins. Added connector integrations and Scheduled Tasks in 2026, and ships a desktop app alongside the web product. It relies on external frontier models, primarily American ones, under the hood.

In practice

A researcher gives Manus a one sentence brief to compare the top ten agent platforms on pricing and build a spreadsheet. It browses each site, extracts the data, runs code, and returns the finished sheet without further input.

A builder asks for a working prototype. Manus 1.5 generates a full web app with backend logic, a database, and authentication from a single prompt.

An analyst needs a deep dossier. Wide Research spins up many sub agents in parallel across sources and assembles a cited report while the user is away.

Capability coverage

7.5 / 14 capabilities · 54%

Integrations & Tool CallingOperates in a cloud virtual computer with real browsers, terminals, and file systems, plus added connector integrations in 2026, so it calls real tools directly rather than through narrow plugins, ssojet and nocode.mba coverage retrieved 2026-07-08 Full
Workflow OrchestrationTakes a high level goal and independently plans and executes multi step tasks end to end, with Wide Research running many sub agents in parallel across sources, ssojet and theplanettools coverage retrieved 2026-07-08 Full
Knowledge Grounding & RAGPerforms deep multi source web research and cites its sources, grounding answers in live retrieval and uploaded files, though it is not a persistent RAG or knowledge base product, nxcode and ssojet coverage retrieved 2026-07-08 Partial
Human Oversight & GuardrailsYou set the goal and review the finished deliverable and can intervene, but Manus runs autonomously in the background with more autonomy meaning less control and no documented approval gate, nxcode and taskade coverage retrieved 2026-07-08 Partial
Security, Identity & GovernanceThe Team plan adds single sign on with SAML and OIDC plus access controls, but reviewers flag that it operates inside authenticated sessions with China founded provider data risk and no published enterprise SLA, theplanettools and ssojet coverage retrieved 2026-07-08 Partial
Observability & AuditabilityShows its steps and takes screenshots during navigation to verify actions, and the Team plan adds usage analytics, though a formal audit log is not documented, nxcode and spectrumailab coverage retrieved 2026-07-08 Partial
Memory & State PersistenceKeeps working in its cloud environment after you disconnect and supports scheduled tasks, giving session and task persistence, though reviewers note limited shared long term memory across sessions, nxcode and taskade coverage retrieved 2026-07-08 Partial
Deployment & Data ResidencyRuns only in Manus's own cloud sandbox with no self hosted, VPC, or data residency options documented, and China founded provider data risk is flagged, theplanettools coverage retrieved 2026-07-08 Unable to verify
Prebuilt Agents, Templates & PacksOffers shared slide templates on the Team plan and produces varied deliverable types, but is fundamentally one generalist agent rather than a library of prebuilt specialized agents, spectrumailab and taskade coverage retrieved 2026-07-08 Partial
Triggers & Channel CoverageSupports Scheduled Tasks and background execution plus connector integrations across web and desktop, a focused set of triggers rather than a broad multi channel framework, nocode.mba and spectrumailab coverage retrieved 2026-07-08 Partial
Model Flexibility & RoutingOffers user selectable tiered model access through Manus 1.6 Lite, 1.6, and 1.6 Max and relies on external frontier models, a limited model choice rather than open bring your own model routing, ssojet and spectrumailab coverage retrieved 2026-07-08 Partial
APIs, SDKs & MCP ExtensibilityAdds connector integrations and a desktop app that extend how it plugs into other tools, but no robust public developer API, SDK, or MCP surface for building on Manus is documented, nocode.mba coverage retrieved 2026-07-08 Partial
Testing, Debugging & OptimizationNo agent testing, evaluation, or debugging tooling is documented; Manus is an end user agent rather than a developer platform with an eval harness, ssojet coverage retrieved 2026-07-08 Unable to verify
Browser & Computer UseIts core capability: runs inside a cloud Linux style virtual computer driving real web browsers, terminals, and file systems, taking screenshots and using vision to verify actions, ssojet and nxcode coverage retrieved 2026-07-08 Full

Recent platform changes

No recent material changes tracked yet.

Pricing

Free tier with three hundred daily refresh credits; paid Pro from twenty dollars per month, Extended at two hundred dollars per month

credit based subscription; every action consumes credits, complex tasks consume many

Public — partialHigh variable costFree tierTrial available

What is public

The full tier structure is public on the pricing page: free daily credits, Pro at twenty and forty dollars, Extended at two hundred dollars, and Team per seat, with credit allowances stated for each. Exact credit consumption per task is not fixed.

Billing mechanics

Credit based subscription tiers. Free gives daily refresh credits only, paid tiers grant a fixed monthly credit allowance that does not roll over, purchased add on credits carry over only while a paid subscription stays active, and annual billing saves about seventeen percent. Team adds single sign on, analytics, access controls, and shared templates.

Cost watchouts

Every action consumes credits and complex tasks burn through them fast, with a wide research run reported to consume four thousand to ten thousand credits, so real spend is hard to forecast and monthly credits do not roll over.

Variable cost rationale

Pricing is credit based and every action consumes credits, with complex tasks reported to burn four thousand to ten thousand in a single run and monthly credits not rolling over, so cost can escalate fast and unpredictably with usage.

Additional watchouts

Because pricing is credit based and consumption scales with task complexity, heavy or long running agent workflows can escalate cost quickly; budget against expected task volume, not just the sticker tier price.

Sales call required

No — self-serve available

Free / trial

Free plan at zero dollars with three hundred daily refresh credits and access to Manus 1.6 Lite, no credit card required

Lowest paid plan

Pro from twenty dollars per month for four thousand monthly credits, with a forty dollar tier for eight thousand credits

Key ambiguities

How many credits a given real workload will consume, since consumption scales with task complexity and long running agents are the biggest cost driver.

Missing data

No published enterprise tier with a stated SLA; large deals go through direct sales at custom pricing.

Verified 2026-07-08

Contact us

Found a vendor we missed? Have feedback on the index? We'd love to hear from you.