China has made a bold leap in artificial intelligence with the Manus AI agent, a groundbreaking development that has caught the attention of tech leaders worldwide, including Silicon Valley. Launched last week via an exclusive invitation-only preview, Manus is China’s most ambitious move into the rapidly evolving AI agent market.
Unlike conventional AI chatbots, Manus is designed to independently handle complex multi-step tasks with minimal human intervention, setting it apart from previous AI models.
China’s AI Breakthrough: Manus AI’s Unique Edge
Developed by the Chinese startup Butterfly Effect with financial support from Tencent Holdings, Manus AI is attracting global interest due to its ability to bridge the gap between theoretical AI capabilities and real-world applications.
At its core, Manus operates on an innovative multi-model architecture that integrates multiple AI models, leveraging their strengths to achieve superior reasoning and execution.
Autonomous AI Task Execution: A New Era
According to Peak Ji Yichao, co-founder and chief scientist at Butterfly Effect, the Manus AI agent was developed using large language models such as Anthropic’s Claude and fine-tuned versions of Alibaba’s open-source Qwen.
This multi-model structure enables Manus to adapt to specific tasks, selecting the most suitable AI components for efficient execution. CNN Business reports that Manus can perform advanced tasks, including screening job applicants, building websites, and even compiling reports based on specific investment criteria.
Hands-On Testing: Manus AI in Real-World Scenarios
To assess its practical effectiveness, MIT Technology Review conducted an extensive test across three key tasks:
- Curating journalist contact lists
- Conducting real estate searches with intricate parameters
- Identifying top candidates for the Innovators Under 35 program
Reviewer Caiwei Chen described the experience as akin to working with a highly intelligent and efficient intern. While Manus occasionally misinterpreted instructions or expedited tasks at the expense of accuracy, it displayed remarkable adaptability and learning ability, improving significantly with user feedback and detailed instructions.
One standout feature is its “Manus’s Computer” interface, which allows users to monitor its decision-making process in real time. This transparent AI workflow provides greater user control while automating complex operations.
Technical Challenges: Scaling Manus AI
Despite its impressive capabilities, Manus AI is still facing technical limitations. MIT Technology Review reported frequent system crashes and timeout errors during extended sessions, with error messages hinting at high server load.
The demand for Manus has surged, but less than 1% of waitlisted users have received invite codes. The official Manus Discord server has grown to over 186,000 members, highlighting the immense interest in this next-generation AI agent.
Furthermore, operating costs remain relatively affordable, with each task costing around $2, according to Chinese tech publication 36Kr.
Strategic Partnership with Alibaba Cloud
To enhance performance and accessibility, Manus AI’s developers have partnered with Alibaba Cloud. A recent South China Morning Post report confirmed that Manus will integrate Alibaba’s Qwen models, making the AI agent more accessible to Chinese users on domestic computing platforms.
While the exact timeline for implementation remains unclear, this partnership aligns with Alibaba’s broader advancements in AI foundation models.
Alibaba’s AI Model Innovation
Alongside the Manus AI launch, Alibaba has introduced QwQ-32B, a new reasoning model designed to rival OpenAI’s o1-mini and the DeepSeek R1 model.
According to CNN Business, QwQ-32B surpasses OpenAI’s o1-mini while maintaining a significantly lower parameter count of 32 billion, compared to DeepSeek’s 671 billion parameters. This suggests that Alibaba has made efficiency breakthroughs, reducing computational requirements while maintaining strong performance.
China’s AI Strategy: A Global Powerhouse
The rapid evolution of Manus AI and Alibaba’s foundation models reflects China’s strategic commitment to AI leadership. The Chinese government has prioritized AI as a key industry, alongside quantum computing and robotics, with substantial investment pouring into research and infrastructure.
Alibaba alone has pledged 380 billion yuan ($52.4 billion) over the next three years toward AI and cloud computing, a sum exceeding its total investment in the previous decade.
As MIT Technology Review’s Caiwei Chen notes, China is not just following Western AI trends but is actively shaping the future of autonomous AI agents through original innovations and strategic investments.
Rather than mirroring Western AI advancements, China is developing distinct AI models and agentic architectures optimized for diverse applications and cultural contexts. This suggests the future will see multiple AI paradigms coexisting, each designed to meet different global needs.