Alibaba, the latest response!

【Editor’s Note】Alibaba CEO Responds to News That Qwen Core Lead Lin Junyang Resigns: “If Technology Doesn’t Move Forward, It Falls Behind”; Company Will Continue to Attract Outstanding Talent

China Fund Reporter Cao Wenjing

On the morning of March 5, Alibaba CEO Wu Yongming responded internally via an email to the news that Lin Junyang, a core leader behind Alibaba’s Qwen, has resigned.

Wu Yongming said the company has decided to approve Lin Junyang’s resignation and thanked Lin Junyang for his efforts in the role. “Jingren will continue to lead the Tongyi Laboratory in pushing forward subsequent work. ‘The company will establish a foundation model support team, coordinated jointly by me, Jingren, and Fan Yu, to support the group’s foundation model development with resources.’”

Wu Yongming pointed out, “If technology doesn’t move forward, it falls behind.” “Developing large foundation models is our key strategy for the future. While we continue to adhere to an open-source model strategy, we will continuously increase R&D investment in the field of artificial intelligence and step up efforts to attract outstanding talent. Let’s do our best together.”

An Alibaba insider told the reporter that the company has not changed Qwen’s open-source strategy, and it has not used commercial goal assessments such as daily active users to evaluate the foundation model team. Lin Junyang’s resignation is unrelated to these rumors. The actual situation is that, as Qwen moved from improving foundation models to becoming part of the group’s overall strategy, the company believes it needs to recruit more top technical experts to increase the talent density of the foundation model team. This process involves adjustments to the scope of authority and responsibilities within Lin Junyang’s role. Lin Junyang did not accept this, and therefore submitted his resignation.

In the early hours of March 4, Lin Junyang, the core leader of Alibaba’s Qwen large model, posted on social media saying, “me stepping down. bye my beloved qwen.（I’m stepping down. Goodbye to my beloved Qwen）,” which drew public attention.

That afternoon, Lin Junyang responded in a post on social media: “Sorry, everyone. I’m not going to reply to messages or calls today. I really need to rest. Qwen brothers, continue working as originally arranged—no problem.”

It is worth noting that before stepping down, Alibaba’s Qwen had just released the Qwen 3.5 small-size model series. After the model was released, it quickly attracted likes and comments from Elon Musk on social media, saying that Alibaba’s Qwen model has “impressive intelligence density.” Lin Junyang thanked Musk for the like on the social platform.

Qwen is Alibaba’s personal AI assistant application launched in China on November 13, 2025. Developed based on the Qwen large models, it is Alibaba’s flagship AI application for consumers. Lin Junyang is the core lead behind Qwen and also Alibaba’s youngest P10-level technical leader.

Lin Junyang is an important leader involved in Qwen models. Public information shows that Lin Junyang was born in 1993. He studied computer science as an undergraduate at Peking University, and in the master’s program he completed his studies at the School of Foreign Languages at Peking University. His field of study is linguistics and applied linguistics. His research areas include natural language processing and multimodal representation learning. He has published multiple papers at top conferences such as NeurIPS, ICML, and ACL.

After graduating in 2019, Lin Junyang joined Alibaba’s DAMO Academy, serving as a senior algorithm engineer. He took part in the development of the ultra-large-scale pretraining model M6 and the general unified multimodal pretraining model OFA.

At the end of 2022, Alibaba integrated the AI teams for language, vision, and more from DAMO Academy into Alibaba Cloud and established the Tongyi Laboratory. Lin Junyang was appointed as the technical lead for the Tongyi Qwen series of large models. In 2023, Lin Junyang drove the release of the Tongyi Qwen large models, whose architecture integrates multimodal technical achievements.

In 2024, Lin Junyang led the open-source work for the Qwen series of models, covering different parameter-scale versions such as 7B, 14B, and 72B. In 2025, it launched its flagship model Qwen3-Max, with a parameter scale of over one trillion and using 36T of pretraining data; in evaluations such as GPQA, it surpassed mainstream international models from the same period. In the same year, it also formed a robotics and embodied intelligence research team, promoting multimodal intelligent agents to extend into real-world applications.