Alibaba Group Holding launches RynnBrain to teach robots physical-world perception and manipulation

Alibaba rolls out RynnBrain to teach robots about the physical world

RynnBrain aims to give robots robust "world models"

Alibaba Group Holding is unveiling RynnBrain, an artificial intelligence model from its DAMO Academy that helps robots perceive and interact with the physical world, the company says. A demo video shows a robot recognising fruit and placing it into a basket, illustrating the model’s combined perception, manipulation and motion-planning capabilities. Alibaba positions RynnBrain as part of a push beyond cloud and e‑commerce into so‑called physical AI, where systems must understand objects, their relationships and how to move within real environments.

The model targets "world models" that integrate object understanding with movement control, a technical area attracting major competitors including Nvidia, Google DeepMind and Tesla. Alibaba links RynnBrain to the momentum of its Qwen family of large AI models and is rolling the new system out as open source to accelerate developer adoption and ecosystem growth. By making the model freely available, the company seeks to foster experimentation by researchers and robotics firms while accelerating commercial use in logistics, warehouses, manufacturing and consumer robotics.

Alibaba frames RynnBrain as both a research milestone and a commercial building block for practical deployments where perception and manipulation are critical. The company highlights potential use cases such as automated sorting in supply chains, robotic picking in distribution centres and collaborative factory robots that handle diverse objects. The announcement signals a strategic diversification for Alibaba as it leverages its AI research arm to pursue hardware‑adjacent applications and compete in the emerging market for integrated perception‑to‑action systems.

Industry context and competition

The launch comes amid heightened global competition in humanoid and robotics AI, with firms in China and the U.S. racing to scale production and field more capable systems. Nvidia and Google DeepMind are developing analogous "world" and robotics models, and Tesla continues to push its Optimus humanoid project, making the space a focal point for major cloud, chip and AI developers.

Practical rollout and ecosystem play

Alibaba highlights the China International Supply Chain Expo appearance as a venue to demonstrate practical ambitions and to connect RynnBrain with logistics and manufacturing partners. The open‑source approach is consistent with Alibaba’s prior strategy of releasing models publicly to broaden adoption and build a developer ecosystem around model‑driven robotics.