Tsinghua AIR spin-out Benshu AI raises round led by 5Yuan Capital for action-native embodied model

A founder from the first cohort of Tsinghua University’s Institute for AI Industry Research has launched his own embodied-AI company, and a top Chinese venture firm is leading the round.

Benshu AI embodied model concept visual
Benshu AI’s action-native embodied model concept. (Source: Leiphone)

Li Jianxiong, the first doctoral graduate of Tsinghua AIR, founded Benshu AI, with 5Yuan Capital, formerly Matrix Partners China, leading the investment. The company is betting on an action-native embodied model, a system built around physical action as the primitive rather than treating motion as an afterthought to language or vision.

The action-native framing is the differentiator. Most embodied models still bolt a control layer onto a perception or language backbone. Benshu’s thesis is that training from action outward, so the model reasons in terms of what a body should do, produces robots that generalise better to unstructured tasks. The bet rides a wider wave of Chinese start-ups spinning out of top university AI labs and chasing embodied intelligence as the next platform shift.

Benshu AI team or lab scene
Benshu AI’s research setting. (Source: Leiphone)

The backing matters as a signal. 5Yuan Capital is among the most established early-stage investors in Chinese technology, and its lead places Benshu in a crowded but capital-rich field where the winners will be decided less by demo polish than by data, talent and a defensible model architecture.

Editor’s note: This is an adapted translation of the original Leiphone report. It has been trimmed and restructured for readability for an international business audience. The full original (in Chinese) is at https://www.leiphone.com/category/ai/WbD0Tf6P1oA2WDdq.html.

Leave a comment