EN / 中文

Alibaba DAMO’s Rynn Lead Li Xin Departs for Physical AI Startup MolinX

by zhinengchecankao·November 15, 2026

Quantum Bit ROBO has exclusively learned that recently, Li Xin, head of the Rynn series models at Alibaba's DAMO Academy, has left his position to start a business, focusing on foundational models for Physical AI.

The Rynn series models at Alibaba's DAMO Academy are designed for Physical AI. During his tenure at DAMO Academy, Li Xin was responsible for the R&D of the Rynn series models and algorithms.

After leaving DAMO Academy, Li Xin joined the startup company MolinX (Muling Technology), which was established in June this year. The founder is Chen Mingxiu, former head of the embodied platform at Alibaba's DAMO Academy. The two had previously collaborated on multiple model projects.

MolinX focuses on self-evolving foundational models for Physical AI.

From Multimodal Research to Physical AI

Li Xin obtained his bachelor's degree from Sun Yat-sen University and received his PhD in Systems Engineering and Engineering Management from The Chinese University of Hong Kong in 2021.

In the same year he completed his PhD, Li Xin joined Alibaba's DAMO Academy through the "Alibaba Star Top Talent Program".

According to Li Xin's GitHub profile, his research direction expanded from natural language understanding to video multimodality, and then shifted to Physical AI.

A representative work during the video multimodality phase was the DAMO Academy's Video-LLaMA series of models. Li Xin previously studied image and video understanding and also explored joint modeling of audio and visual information.

Li Xin served as the corresponding author for VCD, a work on mitigating visual hallucinations, and as co-corresponding author for Inf-CLIP. These works were selected as Highlights at CVPR 2024 and CVPR 2025, respectively.

After shifting to Physical AI, Li Xin's primary focus became the DAMO Academy's Rynn series models.

The Rynn series revolves around how robots understand environments, take actions, predict consequences, and evaluate execution processes.

Several achievements from the Rynn series have been open-sourced, including the cognitive foundation model RynnBrain, the action models RynnVLA-001/002, the world model RynnWorld, and the value model RynnValue.

In February2026, RynnBrain was open-sourced, targeting first-person perspective understanding, spatial localization, physical reasoning, and task planning. Officially, it was claimed to refresh SOTA (State-of-the-Art) benchmarks across 16 Physical AI evaluations. In July of the same year, the series was expanded to 122B-A10B.

As the algorithm lead, Li Xin's responsibilities included advancing the research roadmap, model development, and coordinating collaboration among PhD students and young researchers.

Starting a Business with Former Colleagues, Focusing on Physical AI

Li Xin's next destination is the startup MolinX.

MolinX's founder, Chen Mingxiu, is the former head of Embodied Intelligence Technology at DAMO Academy and was a founding team member of the smart glasses brand Rokid.

Li Xin and Chen Mingxiu continued their collaborative relationship from their time at DAMO Academy. Public papers show that both frequently appeared together in the author lists of the Rynn series models.

Notably, over the past two months, several heads at DAMO Academy have successively resigned.

In addition to Li Xin and Chen Mingxiu, in October this year, Rong Yu, head of Scientific Intelligence at DAMO Academy, left to found Gezhi Emergence, continuing to explore the AI for Science direction.

Earlier, in September, Zhao Deli, Chief Scientist and Head of the Embodied Intelligence Laboratory at DAMO Academy, was also reported to have resigned, with her next destination undisclosed.

Beyond being superior and subordinate colleagues at DAMO Academy, Zhao Deli and Li Xin both worked at Microsoft Research Asia (Zhao from 2006-2012, Li interned from 2015-2016) and both had affiliations with The Chinese University of Hong Kong (Zhao served as a research assistant from 2008–2013, Li pursued his PhD from 2016-2021).