Unitree Technology's Wang Xingxing: Cumulative G1 humanoid robot production offline reaches about 10,000 to 20,000 units; GPT-6 can directly control robots with zero training

Related topics top start Related topics top end Futures promotion begin Futures promotion end Miaopai begin Miaopai end Video player start Video player end Market chart begin Market chart end

Sina Tech News, October 22 afternoon — Wang Xingxing, founder of Unitree Robotics, delivered a speech at the 2026 World Physical AI Conference, reviewing Unitree's progress over the past year and sharing his views on the development trends of physical AI.

Wang Xingxing revealed that the Unitree G1 humanoid robot is currently the best-selling humanoid robot in the world, with cumulative production of approximately 10,000 to 20,000 units. "Today, other robotics companies around the world have successively launched products similar to it, and some products look almost indistinguishable."

Speaking about the impact of GPT-6's release on the robotics field, Wang Xingxing said Unitree's tests showed that with completely zero training, the model was already able to directly control the robot to complete tasks such as drawing, cleaning paintbrushes, and playing Go; the robot can change colors on its own and clean the paintbrush itself, and "the precision of placing stones is very high." He judged that in the coming years, large-model companies' attention to machine intelligence and the robotics industry will continue to increase, and "at the fastest, in one to two years, or at the slowest, in two to three years, the industry is expected to see more progress."

Wang Xingxing proposed that in the future, when robots complete about 80% of tasks, they should need only voice and language instructions to do so — "at the fastest, there is a chance of achieving this next year; to be safer, it may take until the year after." He pointed out that what has not yet been well solved is mainly the interaction between robots and real objects, as well as the alignment between AI model inputs and outputs and real robots, much of which falls under engineering and technical problems.

As for the key directions for large-scale deployment of physical AI, Wang Xingxing believes there are three: first, improving the large-scale production and manufacturing capacity of robot hardware — "we currently produce roughly tens of thousands of robots a year, as well as hundreds of thousands of joint motors and modules, but to achieve truly large-scale adoption, current capacity is far from enough"; second, promoting independent research and development of physical AI models and their industrial deployment — Unitree will increase its own R&D investment while cooperating with large-model companies at home and abroad; third, building larger-scale server clusters and computing power clusters, and possessing the corresponding computing power chip production and manufacturing capacity.

Wang Xingxing also revealed that Unitree is developing "Unitree Rehearsal," a self-evolution system for physical AI robots that lets AI models collect papers on their own, write code, train in simulated environments, and then test in real environments, with test results scored and fed back by the AI models, forming a self-loop. At the end of the speech, Wang Xingxing admitted, "I am really afraid of wasting every day," and said, "Our generation truly has the opportunity to realize superintelligence or AGI, and also the opportunity to let humanoid robots raise productivity to the next level." news_keyword_pub,stock,sz300024&sh688836 iframe below main text iframe below main text General statement General statement QR code at end start QR code at end start

QR code at end start QR code at end start Editor name and work code

Editor name and work code