On March 4th, it was reported that Lin Junyang, the head of Alibaba Group's Qianwen Big Model, announced in a post on social platform X that he would step down from the Qianwen team early this morning.
Lin Junyang did not directly state that he would resign from Alibaba, perhaps only leaving the Alibaba Qianwen team to shift to other business lines.
Chen Cheng, a maintainer of the Magic Community and contributor to Qianwen, shared a comment saying that he was heartbroken. He believed that leaving was not Lin Junyang's choice. Last night (the evening of March 2nd), they jointly released the Qwen3.5 small-sized model series, and it is unimaginable for Qianwen to be without Lin Junyang.
Since March 2nd, Alibaba's Qianwen Big Model has announced a series of significant actions.
On the evening of March 2nd, Alibaba opened sourced four Qwen3.5 small-sized model series: Qwen3.5-0.8B, Qwen3.5-2B, Qwen3.5-4B, and Qwen3.5-9B, which can be deployed on end devices such as laptops. After the model was released, it quickly attracted attention from overseas technology circles. Musk parachuted into the social media platform X and commented on it in a tweet, "Amazing intelligence density." Lin Junyang forwarded and thanked him.
The two consecutive incidents quickly sparked heated discussions among netizens. In Lin Junyang's comment section, many people speculated whether there was a connection between the two, and even some voices believed that Musk had offered a high price invitation to this young AI technology genius.
On March 2nd, Alibaba's Qianwen Big Model announced that it will unify the B-end brand and C-end application brand of the big model into Qianwen, and the name "Tongyi Qianwen" will no longer be used.
Lin Junyang has been with Alibaba for nearly 7 years since 2019. Before stepping down, he served as the head of Alibaba's Qianwen Big Model, a core member of Alibaba Group's Tongyi Laboratory, and the youngest P10 level technical leader at Alibaba.
He was born in 1993 and graduated with a bachelor's degree in Computer Science from Peking University, and a master's degree in Foreign Linguistics and Applied Linguistics from the School of Foreign Languages at Peking University.
During his master's degree, Lin Junyang was involved in projects such as multidimensional text classification and attention mechanism modeling. This experience led him to try to transform language theory into a computable framework.
After graduating with a master's degree, he joined the Alibaba Damo Institute Intelligent Computing Laboratory in 2019 as an algorithm expert and participated in the development of the Alibaba M6 multimodal pre training model.
In 2022, multiple research teams from Alibaba Cloud reorganized and established the Tongyi Laboratory. Lin Junyang transferred from Alibaba Cloud to Tongyi Laboratory and was officially appointed as the technical leader of the Qianwen series large model.
During the years under Lin Junyang's leadership, Alibaba's Qianwen model gradually became the world's strongest open-source model, occupying a leading position in the battle of AI gods. On April 29, 2025, Alibaba's open-source Qwen3 series models surpassed the top global open-source models of the same period in multiple authoritative evaluations, officially topping the list of the world's strongest open-source models. In January 2026, the latest data from Hugging Face, the world's largest AI open source community, showed that the number of Alibaba Qianwen derivative models exceeded 200000, and the download volume of Qianwen series models exceeded 1 billion, firmly ranking first in the world for open source large models.
Last October, he also announced on X that he had formed a robot and embodied intelligence team within Qianwen.
As the leader of the Alibaba Qianwen team, Lin Junyang's resignation may mean that the Qianwen big model team will usher in a critical moment of handover and strategic shift.
According to incomplete statistics, several technical executives from Alibaba's big model team have resigned since 2024.
In December 2025, Huang Fei, the former head of natural language processing (NLP) at Tongyi Laboratory, was exposed to have resigned; In April 2025, Bo Liefeng, the former head of Tongyi Lab's application vision team, left his post. He once led the "King of All Dance", "Terra Cotta Warriors Dance" and other popular apps, and was exposed to join Tencent's hybrid model team after leaving; In February 2025, Yan Zhijie, the former head of the Tongyi Laboratory's voice team, resigned. In August 2025, he joined JD Exploration Research Institute as the head of the voice laboratory; In July 2024, Zhou Chang, a technician of the original Alibaba Cloud Tongyi big model team, was exposed to be leaving. In October 2024, he joined ByteDance to take charge of the AI big model work. He was joined by more than a dozen people in his team who had been sued by Alibaba for violating the non competition agreement.
Conclusion: Top talents in big models enter a period of high-frequency mobility
Since last year, major domestic and foreign companies have frequently experienced changes in their core talents. From Meta, which spared no expense in poaching people from Apple, Google, and OpenAI to establish a super intelligent laboratory, to xAI's co-founder and core technical personnel leaving one after another, to the current Alibaba Qianwen core leader Lin Junyang stepping down. The recent talent flow related to big models at home and abroad shows that the competition for top talents in the big model field has become increasingly fierce, and the global AI core talent flow has significantly accelerated. High paying poaching and core team departure have become the norm in the industry. For Alibaba, Lin Junyang is a core figure in the Qianwen Big Model, which may exacerbate industry concerns about the stability of Alibaba's AI talent. The new trends of this talented AI expert will become the focus of industry attention.
