Manus

Sacra estimates that Manus hit $100 million in ARR in December 2025, up from $90 million in August 2025. The company introduced its first paid subscription plans in March 2025, transitioning from a free, invite-only beta model that it operated throughout 2024. The company reached $100M ARR approximately 8 months after launching paid plans.

Manus has a three-tier pricing structure starting at $19 per month for basic plans and reaching $199 per month for premium tiers.

Meta announced its acquisition of Manus in December 2025 at a reported valuation of more than $2 billion. The deal will result in Manus discontinuing operations in China while continuing from its Singapore headquarters. Meta stated there will be no continuing Chinese ownership interests following the acquisition.

Manus raised $75 million in a Series B round led by Benchmark in April 2025, bringing its valuation to approximately $500 million. This followed a $10 million Series A in January 2023, bringing total funding to $85 million. The Benchmark investment was subject to review by the U.S. Treasury Department in May 2025 regarding compliance with 2023 restrictions on investing in Chinese companies.

Investors include Benchmark as the Series B lead, along with Tencent, HongShan (formerly Sequoia China), and ZhenFund from earlier rounds. The company relocated its headquarters from China to Singapore in 2025 as part of its international expansion.

Manus is an autonomous AI agent platform designed to execute complex, multi-step tasks based on natural language instructions. Users provide task descriptions in plain English, and Manus initiates a dedicated cloud virtual machine equipped with tools such as web browsers, code interpreters, office applications, and design engines.

The platform offers two primary modes: a standard chat interface for Q&A interactions and an agent mode that activates the full autonomous execution environment. For example, when tasked with analyzing 100 sneakers and creating a sortable spreadsheet, Manus decomposes the objective into discrete steps, autonomously browses relevant websites, extracts pricing data, normalizes product information, generates a CSV file, and delivers the output as both chat results and downloadable files.

The system utilizes Anthropic's Claude language models, augmented with proprietary context engineering techniques to optimize prompt scaffolding and memory management. Successive platform updates have driven meaningful performance gains: Manus 1.5 (October 2025) cut average task completion time from 15 minutes to under 4 minutes, with a 15% internal task-quality gain and 6% higher user satisfaction, while Manus 1.6 Max (December 2025) added a Design View feature and raised user satisfaction by more than 19.2% in double-blind testing.

Users can schedule recurring tasks and run multiple parallel workflows based on their subscription tier. The Wide Research feature (launched July 2025) enables up to 100 parallel sub-agents to operate simultaneously, significantly reducing execution time for research-intensive tasks. Manus also allows users to build and publish mobile apps without requiring Xcode or traditional development tools (launched January 2026), positioning the platform to serve non-developers seeking to create and distribute mobile applications directly through Manus.

Manus has also expanded into enterprise workflows through a partnership with Microsoft: as a launch partner for Microsoft Agent 365 (November 2025), Manus workflows can operate natively within Microsoft 365 after enterprise IT setup, turning the Microsoft ecosystem into a distribution channel rather than a competitive threat.

Sacra estimates that Manus hit $125M in revenue run-rate in December 2025, up from $100M ARR eight months after launch, with more than 20% month-over-month growth since the Manus 1.5 release. The company had 105 employees across Singapore, Tokyo, and San Francisco as of December 2025, with a Paris office planned.

Manus operates a B2C subscription model with usage-based pricing tiers that scale with customer requirements. Individual plans include a Starter tier at $39/month offering 3,900 credits and up to two concurrent tasks, and a Pro tier at $199/month offering 19,900 credits and up to five concurrent tasks, both providing longer context, improved multimodal capabilities on Claude 3.7 Sonnet, and dedicated resources with priority access during peak hours. For teams, Manus offers a Team plan at $39 per seat per month with a five-seat minimum ($195/month total), providing 19,500 pooled credits and access to dedicated infrastructure and beta features.

The company employs a freemium go-to-market strategy, allowing users to access basic functionality with limited usage before transitioning to paid plans. During the beta phase, this approach generated a waitlist exceeding 2 million users prior to monetization.

The business model benefits from low marginal costs after the core infrastructure is established, as additional users primarily generate API calls to third-party language models rather than requiring substantial incremental hardware or labor. Manus is also developing an enterprise SDK to allow other businesses to embed Manus workflows directly into their applications, which could transition the company from a direct subscription model to a developer-focused API business.

OpenAI, Microsoft, Google, and Anthropic integrate agent capabilities directly into their foundation model offerings, leveraging integration advantages and subsidized pricing through their cloud infrastructure. Google's Gemini agents integrate with Workspace and Android, while OpenAI's agent features overlap with Manus's focus on autonomous task execution. Microsoft's Agent 365 platform represents a similar dynamic, and Manus's position as one of its launch partners turns a potential displacement threat into a distribution channel within Microsoft 365 workflows.

These incumbents can offer lower pricing by treating agent development costs as a customer acquisition and retention strategy for their broader cloud platforms. However, enterprise buyers increasingly seek vendor-neutral solutions to mitigate lock-in risks, creating opportunities for independent platforms such as Manus.

Adept's ACT-2 employs computer vision to directly control user interfaces, bypassing the API integrations that Manus uses for task execution. Cognition's Devin targets software engineering tasks, achieving quantifiable performance benchmarks on coding challenges that appeal to development teams.

These specialized competitors focus on specific verticals rather than general-purpose capabilities. Rework.ai, Sweep.ai, and similar coding-oriented agents integrate into development workflows and CI/CD pipelines, capturing high-value engineering use cases that might otherwise align with general platforms like Manus.

LangChain, AutoGen, and other open-source agent frameworks are commoditizing the orchestration layer underlying autonomous task execution. Organizations can use these tools to build custom agent workflows instead of subscribing to hosted platforms.

While open-source solutions require greater technical expertise for implementation and maintenance, they provide unlimited customization and eliminate ongoing subscription costs. This dynamic exerts pricing pressure on commercial platforms, which must differentiate through ease of use, reliability, and advanced features rather than basic agent functionality.

Manus is developing an agent-as-platform SDK that allows other companies to embed Manus workflows directly into their applications, shifting its focus from a consumer assistant to a developer platform. This approach targets the broader autonomous agent tooling market, which includes enterprise software integration and workflow automation in addition to individual subscriptions.

The platform model enables Manus to generate revenue from every downstream user interacting with partner applications, rather than relying solely on direct subscribers. Companies can build specialized agent experiences on Manus infrastructure and pay revenue-sharing fees, following a model similar to app stores.

The company is piloting domain-specific task packs for finance, merchandising, and legal research. These bundles include custom prompts, specialized tool integrations, and compliance features. Vertical solutions can command higher pricing than general-purpose subscriptions by addressing specific industry workflows that require deep domain expertise.

This vertical strategy allows Manus to compete with specialized software solutions rather than general-purpose AI agents, potentially accessing larger budget allocations for industry-specific tools. It also creates switching costs as customers integrate domain-specific workflows into their operations.

Manus's partnership with Alibaba's Qwen team provides access to advanced multimodal AI models and preferential cloud computing resources. This collaboration supports expansion into physical-world automation, including factory automation, retail applications, and smart home use cases.

With offices in Tokyo and San Francisco, alongside its Singapore headquarters, Manus is positioned to serve multinational enterprises that prioritize consolidated vendors over regional point solutions. This global presence is particularly relevant for large accounts requiring consistent agent capabilities across multiple countries and regulatory environments.

Model dependency: Manus depends extensively on third-party language models from Anthropic and other providers, which introduces both cost exposure and strategic risk. Increases in API pricing, access restrictions, or the introduction of competing agent platforms by these providers could result in margin compression or the loss of access to critical components of Manus's technology stack.

Regulatory scrutiny: The U.S. Treasury Department reviewed Benchmark's April 2025 investment in Manus for compliance with 2023 restrictions on investing in Chinese companies, highlighting ongoing regulatory risks for AI startups with Chinese origins. While Meta's acquisition will eliminate Chinese ownership interests and discontinue China operations, enterprise customers may remain cautious about data sovereignty issues, particularly for sensitive business workflows processed by Manus agents.

Execution complexity: Deploying autonomous agents for real-world tasks introduces liability and reliability risks that are less pronounced in simpler AI applications. As Manus agents integrate with more external systems and manage higher-stakes workflows, the likelihood of costly errors or security breaches increases, which could constrain enterprise adoption.