by Butterfly Effect Pte. Ltd.
Autonomous AI agent that plans and executes multi-step work from one prompt: research, documents, apps and video. The September 2026 release adds Cascade, Automations and Studio. Free tier available.
Choose Manus if you would rather hand off a whole goal than steer a chat. It suits research reports, slide decks, small web apps and short videos, and it keeps working after you close the tab. Expect credit-based billing, so heavy tasks can drain a plan faster than the sticker price suggests.
Leads the GAIA Level-1 benchmark, the highest publicly documented score for any AI agent, combined with async cloud execution that completes tasks after the user closes their browser
GAIA benchmark leadership provided short-term distribution advantage during the Meta era, but Manus is fundamentally a Claude-based wrapper with agentic orchestration, and the moat depends on proprietary training data and execution speed now that Manus operates independently again, not on a proprietary model.
Main trade-off
Credit system burns unpredictably (500-900 credits per complex task, no cost preview before starting), no formal GDPR certification despite SOC 2 and ISO 27001 attestations, and server instability during peak load makes it unreliable for time-sensitive professional workflows
| Plan | Monthly price | What it includes |
|---|---|---|
| Free | Free | Daily refresh credits with a monthly ceiling, Core agent features |
| Pro Standard | $20/mo | 4,000 credits per month, Standard monthly usage |
| Pro Customizable | $40/mo | 8,000 credits per month, Credit amount you set yourself |
| Pro Extended | $200/mo | 40,000 credits per month, Built for sustained heavy use |
| Team | Custom | Per-seat pricing with pooled credits, Admin dashboard with usage stats, Seat and credit selector on manus.im/team |
Pro Standard $20/mo (4,000 credits), Pro Customizable $40/mo (8,000 credits, with a 7-day free trial) and Pro Extended $200/mo (40,000 credits) are listed on manus.im/pricing, each with 300 refresh credits a day, 20 concurrent tasks and 20 scheduled tasks; annual billing takes 17% off. Manus's help center, read on 3 November 2026, gives the same Pro starting prices, puts Free at $0 with 300 refresh credits a day, 1 concurrent task and 2 scheduled automations, and prices Team from $20 per seat a month with a pooled credit pool chosen on manus.im/team. An earlier help article (18 June 2026) capped free refresh credits at 1,500 a month, so confirm the current free allowance in the app. Manus says the pricing page is the final word. Cloud Computer, the dedicated environment, is bought separately at a price shown at purchase. Complex tasks can burn many credits with no preview; compare Genspark and ChatGPT on the same job before committing to a large plan.
| Feature | Free | Pro Standard | Pro Customizable | Pro Extended | Team |
|---|---|---|---|---|---|
| Monthly credits | Daily refresh only (cap 1,500) | 4,000 | 8,000 | 40,000 | Pooled, chosen by seat count |
| Daily refresh credits | Yes | 300 a day | 300 a day | 300 a day | Per team plan |
| Concurrent and scheduled tasks | Limited | 20 of each | 20 of each | 20 of each | Set by the team plan |
| Wide Research | Limited | Yes | Yes | Yes | Yes |
| Annual billing discount | No | 17% | 17% | 17% | Yes |
Manus is an autonomous AI agent from Singapore's Butterfly Effect that plans and runs multi-step tasks from one prompt inside a cloud sandbox. A September 2026 release adds the Cascade agent framework, event-triggered Automations and a Studio workspace, while the three paid Pro plans keep their earlier prices.
Autonomous End-to-End Task Execution
Manus takes a single natural-language goal and independently plans, executes and delivers a finished result such as an 8-page market analysis with source citations in about 15 minutes.
Cascade Agent Framework
Announced in September 2026, Cascade keeps a project light and brings in specialised capabilities only when the work needs them, and the vendor reports lower token use in one tested configuration.
Event-Triggered Automations
Beyond scheduled tasks, Automations start work when a new email, calendar event, Slack message, Notion update or ad-performance change occurs in a connected service.
Manus Studio Video Editor
Studio hands you the timeline behind a generated 30 to 60 second video so you can swap the song, drop in your own footage and re-export, with an Alchemy mode for creative direction.
Game Dev with Cloud Computer Multiplayer
Game Dev starts from playable templates and lets you edit code, assets and scenes; multiplayer runs on a purchasable Cloud Computer that stays up when your laptop closes.
Remote Control and Computer Use
From your phone you can ask Manus to find a file on your desktop and watch it work live, inside a workspace limited to the files, browser and apps you approved.
Async Cloud Operation
Tasks continue in Manus's cloud after you close the browser or app and send a notification when complete, so long jobs need no supervision.
Wide Research Mode
Runs parallel searches across dozens of sources at once to produce structured, cited reports, with task scaling included on every Pro plan.
Mail Manus Email Automation
Forward emails to Manus to trigger actions such as resume screening or meeting prep, using configurable prompt rules.
Most users expect these and they are not currently supported.
| Tool | Price from | Free plan | Runs on | Rating |
|---|---|---|---|---|
| $20/mo | Yes | Web · iOS · Android · Mac · Windows | 3.7 | |
| -- | -- | -- | -- | |
| $24.99/mo | Yes | Web · iOS · Android · Mac · Windows | 3.6 | |
| $25/mo | Yes | Web | 4.4 |
Proprietary
Pro Standard is the entry paid plan at $20 monthly, giving 4,000 credits. Pro Customizable steps up to $40 monthly and 8,000 credits, and Pro Extended to $200 monthly and 40,000 credits. Annual billing saves 17%. Team billing runs per seat, and Cloud Computer is a separate purchase.
Yes. Free accounts receive a daily credit refresh with a monthly ceiling of 1,500 credits, according to Manus's help center. The pricing page shows only the paid cards, so check the app for today's exact free allowance before planning a project around it.
Manus 2.0, announced 28 September 2026, introduces the Cascade agent framework, event-triggered Automations, buyable Cloud Computer environments and a Studio desktop workspace with a Video Editor and Game Dev. The vendor says Cascade cut tokens by 23.2% in one tested configuration; that figure is self-reported.
For general autonomous research Hokai lists Genspark and ChatGPT, for coding agents Devin and Claude Code, for app building Lovable and Replit, and for self-hosting OpenClaw. Which one wins depends on whether the output is a report, a prototype or production code.
Pick Manus for a managed cloud agent with web, desktop and mobile apps and no infrastructure to run. Pick OpenClaw when you need an MIT-licensed, self-hosted agent whose code you can audit or modify, which Manus's closed-source service does not allow.
Manus is a general-purpose autonomous AI agent built by Singapore-based Butterfly Effect Pte. Ltd., launched on 6 March 2025. Unlike a chatbot, it takes a goal such as "build a competitor analysis report", plans the work, researches, writes and hands back a finished file, running asynchronously in the cloud so it keeps going after you close the tab. The company reported crossing $100M in annualized revenue within its first year. Meta announced a $2 billion deal to acquire Manus in December 2025, but was forced to call it off; Manus resumed independent operations in August 2026. On 8 October 2026 parent Butterfly Effect said it had raised more than $500 million, led by Boyu Capital and IDG Capital with existing shareholders Tencent, HSG and ZhenFund taking part. The company did not disclose a valuation, and a Hong Kong listing is reportedly under consideration (TechCrunch, 8 October 2026).
On 28 September 2026 Manus announced Manus 2.0, which it describes as a new architecture rather than a version update. The core is Cascade, its in-house agent framework; in one tested configuration the vendor says Cascade used 23.2% fewer tokens, finished 28.2% faster and cost 32% less than the previous system (self-reported, one configuration). New pieces include Automations, which start a task when something happens in a connected service such as a new email, a calendar event, a Slack message or a Notion update; Cloud Computer, a dedicated environment you buy for a project that needs an always-on server; and Manus Studio, the upgraded desktop app, with a Video Editor (a timeline for 30 to 60 second product ads and tutorials), Game Dev (playable templates, with multiplayer hosted on a Cloud Computer) and Remote Control with Computer Use. Cue, a separate personal-agent app, launched in invite-only early access and is not part of Manus itself.
Pricing is credit based, with three Pro tiers plus a per-seat Team plan; the exact rates and limits sit in the pricing notes and FAQ below. Manus fits solo researchers, product teams prototyping web apps, freelancers producing client deliverables and small agencies running content pipelines. For a broader shortlist see agentic AI tools, or let Smart Match narrow it to your job.
Where the alternatives win: Devin, Claude Code and Replit are the safer picks when the output is production code that a team must review, because Manus decides architecture on its own and its output often needs revision. Lovable is narrower but more predictable for a first web app. OpenClaw is the choice if you must self-host under an open licence, and Genspark, Claude or ChatGPT cover research-style work from a familiar interface. The new features are days old at the time of this review, so independent testing of Studio, Automations and Cloud Computer is thin. Hokai's own autonomy test of seven agents and the guide on fully autonomous versus assisted tools explain why supervision still matters.
Product facts on this page (pricing, plans, integrations, compliance) are read from Butterfly Effect Pte. Ltd.'s own pages and documentation. The verdict is Hokai's reading of them, not the vendor's claim.