Alibaba’s Qwen tech lead steps down after major AI push

Alibaba’s Qwen AI project has unexpectedly lost one of its most visible technical leaders, Junyang Lin, just one day after the launch of its new Qwen 3.5 small models. Lin, described by colleagues as a central figure in the project, announced he was “stepping down” in a post on X, offering no further explanation. The abrupt departure from one of China’s most prominent open-weight AI efforts raises immediate questions about the project's leadership and stability as global AI competition intensifies.

The announcement followed Monday’s release of Alibaba's Qwen 3.5 Small Model series, a family of four native multimodal systems ranging from 0.8 billion to 9 billion parameters. Designed for applications like on-device deployment and lightweight agents, the models garnered positive attention in the developer community, with Elon Musk noting their “impressive intelligence density.” Lin had been part of the Qwen team since its public launch in April 2023, and his departure came as the project was actively pushing new releases.

The move prompted strong reactions from industry partners and team members, who highlighted Lin's critical role in connecting Qwen with the global developer community. Wenting Zhao, a research scientist on the team, called it “the end of an era,” while Tiezhen Wang of Hugging Face described it as an “immense loss.” The circumstances remain unclear, but a post from project contributor Chen Cheng addressed to Lin, stating, “I know leaving wasn’t your choice,” suggests the decision may not have been voluntary. Alibaba has not commented on the reasons for the departure or the team's leadership structure moving forward.

The departure of a key technical leader responsible for global community engagement, immediately following a successful product launch and amid hints of internal discord, signals potential instability that could disrupt the momentum of a leading open-source AI project.