eWEEK content material and product suggestions are editorially impartial. We could earn money whenever you click on on hyperlinks to our companions. Learn More.
A growth workforce primarily based in China lately unveiled an autonomous AI agent often known as Manus. While lots of the particulars stay scarce, their answer has reportedly outperformed OpenAI – one in all their major opponents – in early benchmarks.
What is an AI agent?
Designed to make knowledgeable selections, carry out primary duties, and be taught primarily based on their prior experiences and interactions, AI brokers symbolize the following technology of digital assistants.
Manus is an autonomous AI agent. Whereas many present AI fashions require human interplay through text-based chat or voice command, Manus features with out the necessity for step-by-step directions. Capable of working independently, a few of Manus’ preliminary capabilities embrace sourcing B2B suppliers, mapping potential prospects, creating instructional supplies, and journey planning.
While AI brokers aren’t precisely new, the rise of huge language fashions (LLMs) has boosted their recognition. When utilized in tandem, AI brokers and LLMs make it simpler to work together with AI and obtain particular goals.
Analyzing the early benchmarks
Known as Manus AI, the Chinese-based growth workforce behind the brand new AI agent posted an introductory video on YouTube to announce their newest innovation. Not solely did it cowl some typical use circumstances, corresponding to resume screening, actual property property analysis, and inventory evaluation, nevertheless it additionally highlighted Manus’ leads to early benchmark checks.
They used GAIA, a standard benchmarking system for AI assistants and different generative AI instruments, to check Manus’ means in fixing real-word issues. When in comparison with earlier benchmarks that have been as soon as thought-about “state-of-the-art” (SOTA), Manus scored greater on all three problem ranges.
But the checks didn’t cease there. The Manus AI workforce additionally measured Manus’ efficiency instantly in opposition to OpenAI, and Manus bested these numbers, too.
- Level 1: Manus (86.5%) / OpenAI (74.3%) / Previous SOTA (67.9%).
- Level 2: Manus (70.1%) / OpenAI (69.1%) / Previous SOTA (67.4%).
- Level 3: Manus (57.7%) / OpenAI (47.6%) / Previous SOTA (42.3%).
Their preliminary benchmarks are promising, however Manus is presently solely out there by way of an invitation-only preview. An precise launch date has not been introduced, and it’s unclear when the brand new AI agent shall be out there to most of the people.
Ramping up the AI fashions competitors
Competition within the present AI panorama is growing each day. Between Manus and a number of other different forms of AI fashions which have been launched in 2025, the yr is already shaping as much as be a pivotal time within the growth of next-gen AI.
