Qwen AI Team Sees Technical Lead and Core Contributors Depart Amid Restructuring

Junyang Lin, the technical lead for Alibaba's Qwen large language model, publicly announced his departure on X (formerly Twitter) on March 3, 2026, shortly after the release of the Qwen3.5 small model. At least two other core contributors, Binyuan Hui and Kaixin Li, also confirmed their exits. The departures have sparked significant discussion within the open-source AI community.

Portrait of Junyang Lin, former technical lead for Alibaba's Qwen large language model.

Lin, 32, was Alibaba's youngest P10-level employee and a key architect of the Qwen series after joining Damo Academy in 2019. His tweet, stating "me stepping down. bye my beloved qwen," was followed 14 minutes later by a post from team member Chen Cheng, which included the phrase "I know leaving wasn't your choice." This suggests the departures may not have been voluntary.

Key Personnel Departures

Beyond Junyang Lin, two other individuals confirmed their exits:

• Binyuan Hui: A co-author of the Qwen technical report and initiator of the OpenDevin project, Hui's research focuses on autonomous programming models. Yuchen Jin, co-founder of Hyperbolic Labs, tweeted that Hui "also left Qwen," and Hui's X bio now states "Formerly MTS at Qwen." His personal homepage, however, still lists him as a staff research scientist at Alibaba Qwen Team, which may be an outdated detail.

• Kaixin Li: A Ph.D. student at the National University of Singapore and a core contributor to Qwen3.5, VL, and Coder. Li posted "Signing off from @Alibaba_Qwen… Onwards and upwards!" He also indicated that a potential Qwen base in Singapore, which he saw as a reason to stay, might now be shelved following Lin's departure.

Three digital silhouettes dissolving into code, symbolizing key personnel departing an AI project.

These individuals were central to Qwen's development, from model training to its open-source ecosystem.

Circumstances of Departure

The phrase "leaving wasn't your choice" from Chen Cheng is the strongest indication that the departures were involuntary. However, Alibaba has not issued an official explanation.

Community speculation points to a restructuring by Alibaba Cloud's senior management, potentially involving a new technical lead for the Qwen team. This could have altered Junyang Lin's reporting structure or influence, leading to disagreements and his eventual departure. Some on X have named Hao Zhou, with a background at Google DeepMind and involvement in the Gemini models, as the new leader. While Zhou's research interests align with large language models and multimodal AI, Alibaba has not officially announced this appointment, nor has Zhou confirmed it.

Abstract neural network with diverging paths, symbolizing corporate restructuring and strategic shifts.

Another theory circulating on platforms like Linux.do and Reddit suggests that Alibaba imposed daily active user (DAU) performance targets on the Qwen team, shifting focus towards consumer-facing products. This emphasis on product metrics reportedly conflicted with the team's open-source-first approach. This aligns with Alibaba's launch of the consumer-facing Qwen App in November 2025 and reports that Alibaba lagged competitors in user scale. However, there is no direct evidence linking DAU pressure to the leadership change.

Timing and Impact on Qwen

The departures occurred less than 24 hours after the launch of four new Qwen3.5 small models on March 2. This timing suggests that the personnel changes were likely decided in advance, awaiting the completion of the scheduled releases.

In the short term, Qwen's GitHub repository continues to be updated, and model weights were released as planned, indicating no immediate halt to open-sourcing efforts. However, the mid-term outlook for Qwen's open-source ecosystem faces uncertainty. The simultaneous exit of the technical lead and core contributors could affect the continuity of long-term projects such as large-scale reinforcement learning and global language coverage.

The community's reaction reflects concerns, with many developers expressing sentiments like "Qwen is nothing without its people." This could impact developer confidence in future versions and enterprises' risk assessment when considering Qwen models. A new leader, potentially from DeepMind, might introduce different technical perspectives, but rebuilding team cohesion and trust within the open-source community will require time.

Oak tree shedding leaves while new saplings grow, symbolizing leadership transition and future growth.

Future Outlook

Junyang Lin's current plans are not publicly known. His GitHub page still lists him as a "Core maintainer of Qwen Team & OpenDevin," and his personal homepage retains an Alibaba email address, which may be due to ongoing departure procedures or a lack of updates. Speculation about his next move, including starting a new company or joining competitors, remains unconfirmed.

This situation differs from a wave of P10-level attrition at Tongyi Lab in 2024, which was primarily driven by high-paying recruitment from companies like ByteDance and Tencent. The current departures appear to stem from internal adjustments rather than external poaching.

A detail noted by Kaixin Li suggests that plans for Qwen to establish an international base in Singapore might now be shelved. Alibaba has not yet issued an official statement regarding the departures or the future leadership of the Qwen team. Future developments, including the release schedule and quality of upcoming Qwen versions, the public appearance of any new leader, and Junyang Lin's next professional steps, will provide further clarity.