Manus

全球首款通用型AI Agent，能够将想法转化为行动，自主完成复杂任务

Agent

https://manus.im/

Visit Website

In-depth Report

Manus is a general-purpose AI Agent (autonomous agent) product released by Chinese AI startup Monica.im (Butterfly Effect) in March 2025. It is positioned as the world's first AI agent capable of "autonomous decision-making in all scenarios." This product can directly transform users' ideas into actual results, and independently complete the entire closed-loop operation from task planning to result delivery. The release of Manus has attracted widespread attention in the technology circle. The internal beta invitation code has been sold to tens of thousands of yuan, and it has also been accompanied by many controversies and doubts.
The development team behind Manus is Monica.im, an AI startup led by Chinese serial entrepreneur Xiao Hong. Xiao Hong is a founder with rich entrepreneurial experience. His entrepreneurial journey can be traced back to his attempt at a campus social project in 2015. Later, in 2020, he accumulated millions of users with the overseas market product ChatGPT for Google. In 2022, the team will take advantage of the wave of large models to transform the field of AI, focusing on the development of AI assistants and intelligent agent products. In January 2025, the company officially changed its name to Butterfly Effect, and released the Manus product in March of the same year. The company has received multiple rounds of financing and has certain capital support in the field of AI applications. Manus was officially released in the early morning of March 6, 2025, using an internal beta invitation system. Just three days after its release, the number of people waiting to register exceeded 200,000, sparking widespread discussion in the technology circle. On May 13, 2025, Manus officially opened for registration, and all users can use it without an invitation code. On March 28, 2025, Manus announced its charging plan, marking its transition from the free testing phase to commercial operation. In June 2025, the team launched a free chat mode and Playbook template library to deal with the dilemma of users “not knowing how to use it”. The core positioning of Manus is a "universal AI agent". Unlike traditional AI assistants that only provide answers, Manus can independently plan, execute and deliver complete work products. Officials describe it as a "virtual colleague with his or her own computer" capable of planning, executing and delivering complete work results from start to finish. This positioning differentiates it from traditional chatbots and AI assistants, and is regarded as a milestone product in the field of AI Agent.
Manus has the following four core features, which give it a unique competitive advantage in the AI Agent market. Autonomous: Manus is able to plan and execute multi-step tasks autonomously, without the need for continuous user supervision and manual intervention. Users only need to describe their goals, and Manus will automatically break down the task into specific steps and execute them one by one. Versatile: Manus can handle many types of work tasks, including but not limited to data analysis, code writing, content creation, document processing, research analysis, etc., truly achieving the universal goal of "one Agent completing various tasks". Multimodal capability (Multimodal): Manus can understand content in multiple formats such as text, images, and documents, and can respond in a human-like manner to adapt to the needs of different scenarios. Privacy protection (Private): The basic version can be used without an account, and the conversation content will not be stored on the server to meet users' needs for privacy and security. According to official and user tests, specific task scenarios that Manus can handle include: research and analysis categories such as comparing cloud service providers, analyzing key points of a 40-page PDF document, summarizing research reports in specific fields, conducting market research and competitive product analysis; data processing categories such as cleaning and visualizing messy spreadsheets, generating data charts and insights, and conducting financial data analysis and modeling; content creation Categories include writing blog posts and product copywriting, generating creative brand names and domain name suggestions, and making presentation slides; programming and development categories include writing code scripts such as Python, debugging program errors and tracebacks, and developing simple desktop applications and websites; productivity tools include making travel plans and generating detailed strategies, drafting professional business emails, and managing schedules and task arrangements. The usage process of Manus is very simple, and users can complete complex tasks in only three steps: the first step is to describe the task, and the user inputs the target requirements in natural language, and can attach relevant documents if necessary; the second step is to perform autonomously, and Manus automatically breaks down the task into multiple sub-steps, conducts necessary research and information collection, and starts execution; the third step is to review and iterate, and the user obtains the final result, which can be optimized and adjusted through subsequent messages. Manus supports direct access from the web and also provides mobile and desktop applications.
Manus adopts a points-based pricing model, and users can choose different levels of packages according to their own needs. The package system includes: Free (free version) provides limited points for free, 1 concurrent task, suitable for those who want to explore and experience; Starter is US$39 per month, provides 3900 points, 2 concurrent tasks, and is suitable for individual users; Pro is US$199 per month, provides 19900 points, 5 concurrent tasks, and is suitable for professional users; Team has customized prices, customized points, and customized concurrency, suitable for corporate teams. In terms of points mechanism, points consumption depends on the complexity of the task and resource requirements; package points are reset monthly, but purchased additional points never expire; users can purchase additional point packages at any time; Team packages provide a shared point pool. Manus's commercialization path reflects the typical AI product development model: first, it accumulates user reputation and topic popularity through a free invitation system, then opens registration in a timely manner to expand the user base, and finally launches paid packages to realize commercial monetization. While this strategy brings cash flow, it has also been criticized by some users for "liquidating too quickly".
In the early days of the release of Manus, user reviews were polarized. According to public opinion analysis, about 70% of users gave positive reviews in the initial stage, mainly focusing on the following aspects: innovative breakthroughs in "execution-level AI" capabilities; impressive multi-tasking capabilities; and stunning product concepts and demonstration effects. At the same time, there are also 30% negative reviews. The main problems include: it takes too long to complete the task and the user experience is unstable; the task success rate is lower than expected; some complex tasks cannot be completed smoothly. After registration opened on May 13, 2025, user reviews became more diverse. Some users said that the product has powerful functions and is particularly suitable for scenarios where repetitive tasks need to be automated. Some users also reported that the free version has more functional restrictions and is less willing to pay. Some users pointed out that it is difficult to get started with the product, and "not knowing how to use it" has become a common problem. In response to user feedback, the Manus team launched the chat mode (Chat) and Playbook template library in June 2025 to lower the user threshold. These improvements have improved the user experience to a certain extent, but product stability and task success rate are still the focus of users' attention.
The release of Manus triggered a strong reaction from the capital market. On the day of the release, A-share intelligent concept stocks collectively hit the daily limit, and concept stocks such as Software 50 ETF, Hande Information, and Tuowei Information rose one after another. Investment institutions generally believe that Manus represents a major breakthrough in the field of AI Agent, and continue to be optimistic about the innovative development of domestic AI large models and applications. Some media call Manus the "GPT moment" in the field of AI Agent, believing that it has opened a new era of AI from "dialogue" to "execution". On a technical level, Manus achieved SOTA (state-of-the-art) results on the GAIA benchmark, excelling at all three difficulty levels. From the perspective of technical architecture, Manus adopts a design similar to a "virtual computer" to allow AI to have an independent operating system environment to perform tasks. This innovative design has been recognized by the industry. However, some technical viewpoints point out that Manus may have "shell" doubts in its engineering implementation, that is, it relies more on existing large model capabilities for engineering packaging rather than underlying algorithm breakthroughs. It is worth noting that Manus has received relatively little attention in overseas markets, in contrast to its popularity in China. Some overseas media and practitioners have reservations about its technological innovation, believing that there is a gap between product promotion and actual capabilities. This phenomenon of “hot at home but cold abroad” has also triggered discussions on the international positioning of products.
Technical authenticity controversy: After the release of Manus, the authenticity of its demonstration video was questioned. Some industry insiders pointed out that the task execution process shown in the video may be artificially manipulated or selectively displayed, and there may be a gap between the actual product capabilities and the demonstration. In addition, there has been a lot of discussion about whether the product is a "shell" of overseas technology, although the team clearly denied this statement. Marketing strategy controversy: The invitation code system adopted by Manus was criticized as "hungry marketing". In the early days of product release, the registration quota was strictly limited, which led to scalpers selling invitation codes on platforms such as Xianyu, and the price was once raised to 50,000 to 100,000 yuan. Although this kind of marketing method effectively created topic popularity, it also caused resentment among users and was accused of "over-marketing". Controversy over the speed of commercialization: The paid package was launched just one month after the product was released, and the paid price ($39-199/month) was relatively high, raising questions about "too fast realization". Some users believe that the product is still in the Beta stage, the service stability and functional integrity have not been fully verified, and the charges at this time are not reasonable enough. Market risk: Competition in the AI Agent market is becoming increasingly fierce, with giants such as OpenAI, Google, and Anthropic all deploying in related fields. As a product of a start-up company, Manus faces multiple challenges such as technology iteration, user retention, and commercialization. In addition, insufficient product experience and stability may also affect user reputation and market performance.
Based on product features and pricing, Manus is mainly suitable for the following groups: individual users include individuals who need to handle daily tasks such as data analysis, document writing, and itinerary planning, AI enthusiasts and followers of the forefront of technology, and freelancers who need automated tools to improve work efficiency; small teams include entrepreneurial teams that need teamwork to complete content creation, market research, and other tasks, and small and micro enterprises with limited budgets but have automation needs; enterprise users include medium and large enterprises that need to automate specific business processes, and traditional enterprises that have AI transformation needs but lack self-research capabilities. New users are recommended to start with the free version and try simple tasks first to understand the product capabilities; use the official Playbook template library to get started quickly; clarify the task goals and provide clear instructions to improve the task success rate. When making payment decisions, it is recommended to choose the appropriate package based on actual usage needs and avoid impulsive subscriptions; pay attention to the point consumption speed, complex tasks may consume a lot of points; you can first purchase additional points to experience high-complexity tasks, and then decide whether to upgrade the package. Suggestions for optimizing usage: It is recommended to execute complex tasks step by step and review intermediate results in a timely manner; use the batch processing function to merge similar tasks; check points usage reports regularly to optimize usage habits.
As the world's first general-purpose AI Agent product, Manus represents an important milestone in the evolution of AI from "dialogue generation" to "autonomous execution". It innovatively realizes autonomous planning and execution of multi-step tasks, achieved excellent results in the GAIA benchmark test, and gained widespread attention from the capital market. However, the controversies since the launch of the product cannot be ignored: questions about the authenticity of the technology, controversy over starvation marketing, and excessive commercialization have all had a certain impact on its brand image. From the perspective of product value, Manus provides users with a complete solution from "thinking" to "action", which is especially suitable for user groups who need to handle repetitive and process-oriented tasks. Its points-based subscription model also provides flexible options. However, improving product stability and lowering the user threshold are still areas that the team needs to continue to optimize. Looking to the future, competition on the AI Agent track will become increasingly fierce. Whether Manus can remain competitive in a market surrounded by giants depends on the balance of its technology iteration capabilities, product experience optimization, and commercialization strategies. For ordinary users, Manus, as a pioneer in the field of AI Agent, is worthy of attention and experience, but it is recommended to view its promotional effects with a rational attitude and make usage decisions based on actual needs.

User Reviews

MiaMiller

—

用了三个月的Manus，不得不说研究能力确实强，但服务器真的经常挂，加载个任务要等半小时。

ocAND

—

199刀一个月的Pro版太贵了，算下来比请个实习生还贵，学生党伤不起。

OliviaParker_2021

—

异步功能真的香，晚上丢个任务，第二天早上起来看结果，效率拉满。

Sophia.Johnson_2021

—

实测做数据分析很给力，输入CSV直接给我生成可视化图表，省了很多功夫。

RArey

—

免费版1000积分根本不够用，做两次深度研究就没了。

Joe447

—

强烈推荐做电商数据分析，竞品调研比人工快多了。

DRper

—

服务器繁忙预警.JPG 每天都要来几次，真的栓Q。

Aiden588

—

Meta收购被否决了，国产AI太难了，希望团队挺住。

JacquelineLong

—

做内容生成的都给我去试试，比ChatGPT香多了。

Ralph_RamosX

—

生成PDF报告的功能很实用，直接能交付，不用自己排版。

Anthony.Anderson369

—

等了两周终于收到邀请码了，感动。

JessicaBennett_Pro205

—

比Cursor的AI能力更全面，但稳定性不如Cursor。

BlocnHub

—

官方文档很详细，小白也能上手。

MrMiguelaVieira_88

—

深度研究能力真的惊艳到我了，_sources引用标注很清楚。

DrAmberSanchez_88

—

199美元/月有点肉痛，但是 productivity 提升了值回票价。

MadisonWilliams_2023

—

不建议做生产级应用，bug太多了，空ZIP文件问题到现在还没修。

Ruth.Green_X

—

数据分析师必备，给它一个Excel 还你一套可视化图表。

MarySanchezQ

—

中国AI的骄傲！

ALong168

—

收费方案出来了，39刀起步，还行吧。

Judy_Price_X

—

研究任务丢给它就能一直跑，不用盯着，懒人神器。

William.CollinsJr

—

讲个笑话：我用Manus帮我写Manus的使用评价

Lawrence60

—

刚升级Team版，企业级功能确实更完善，SSO集成很方便

暖阳218

—

看了创始人访谈，说AI将推动白领工作回归「劳动与运动并重」，这个观点挺有意思

DorisPeterson_77

—

有一说一， Manus的隐私保护做得不错，不需要注册就能用基础功能

BWright_7

—

希望官方能出移动端App，这样外出的时候也能用

Kyle_Perry_99571

—

实测比较：用Manus做旅行规划比ChatGPT更省心，直接生成完整攻略

CynthiaBrown369913

—

薅羊毛失败！免费版积分太少，根本不够用，果断升级Pro

JasonHart_202239

—

作为一个程序员，个人感觉Manus在代码生成方面还有很大提升空间

AVasquez_88

—

看到 Manus 并入 Meta 的消息了，以后会不会变成国际化产品？

流年_2

—

别的不说，Manus的UI设计确实舒服，用起来很顺手

Manus

In-depth Report

User Reviews

Similar Products