Alibaba’s Quon team of AI researchers has been one of the most prolific and well-respected by the international machine learning community – shipping dozens of powerful generalized and specialized generative models starting last summer, most of them completely open source and free.
But now, just 24 hours after shipping the open source Quen3.5 small model series – a release that drew public praise from Elon Musk "effective intelligence density"-The project’s technical architect and several other Quen team members have left the company under unclear circumstances, leading to questions and concerns from around the world about the future direction of the Quen team and its focus on open source.
Junyang’s departure "Justin" Tech chief Lin, who led Quyen from a nascent lab project to a global powerhouse with more than 600 million downloads, along with his two fellow colleagues – staff research scientist Binyuan Hui and trainee Caixin Li – marks an unsettling inflection point for Alibaba Cloud and its role as an international open source AI leader.
These three members of the Quon team announced their departure on X today, although they did not give a reason or say whether it was voluntary or not. VentureBeat contacted Alibaba sources for more information and will update when we get it. Lynn herself signed off with a simple post: "I’m getting down. Goodbye my dear Quen."
While the company is celebrating a technological victory, the sudden exit of its core leadership reveals a deep rift between the researchers who built the models and the corporate hierarchy that is now moving toward aggressive monetization.
Late researchers’ final gift: pocket-sized intelligence
Qwen3.5 small model series (from 0.8B to 9B parameters) represents the final masterstroke "intelligence density" From the founding team.
The models employ a gated DeltaNet hybrid architecture that allows 9b-parameter models to rival the reasoning capabilities of much larger systems.
By using a 3:1 ratio of linear focus to full focus, the models maintain a 262,000-token context window, while remaining efficient enough to run natively on standard laptops and smartphones – even in web browsers.
Lin, a PKU humanities graduate and multilingual, has long advocated "Algorithm-Hardware Co-Design" To overcome computation bottlenecks – a philosophy he detailed at the January 2026 Tsinghua AI Summit.
For the developer community, Qwen3.5 wasn’t just another update; This was a blueprint for "agent inflection," Where models transform from chatbots to autonomous "All-in-one AI worker" Able to navigate UI and execute complex code.
enterprise dilemma
For the more than 90,000 enterprises currently deploying Quon through DingTalk or Alibaba Cloud, the leadership void creates a crisis of confidence.
Many companies moved to Quon because it offered a "third way": Display of a proprietary American model with open weight transparency.
Alibaba has recently consolidated its AI efforts "QUEEN C-END BUSINESS GROUP," Merging its model labs with consumer hardware teams. The goal is clear: to transform Quen from a research project into an operating system for a new era of AI-integrated glasses and rings.
However, the reported appointment of Hao Zhou, a veteran of Google DeepMind’s Gemini team, to lead the Quon team signals a change. "research-first" To "metric driven" Leadership.
Industry analysts, including those cited by InfoWorld, have warned that as Alibaba pushes to meet investor demands for revenue growth, "open" Quen’s open-weight models may become a secondary priority – as we saw with Meta after the disappointing release of its Llama 4 AI model last spring, and the subsequent restructuring of its AI division, the appointment of Scale AI co-founder and CEO Alexander Wang, and the departure of lead researcher Yan LeCun.
Enterprises relying on the Apache 2.0-licensed Qwen model now face the possibility that future flagships – such as the rumored Qwen 3.5-Max – will be locked behind paid, proprietary APIs to drive cloud DAU (Daily Active User) metrics.
Takeaway? If you value Quen’s open source efforts, download and preserve the models while you still can.
"gemini fantasy" Quen’s?
The internal friction at Alibaba mirrors the tensions seen at OpenAI and Google: "Soul" There is often a problem with the machine "scale" Of business. Xinyu Yang, a researcher at rival Chinese AI lab DeepSeek, expressed this sentiment in a candid post on X: "Replace great leaders with non-leaders on Google Gemini driven by DAU metrics. If you evaluate Foundation Model teams like consumer apps, don’t be surprised if the innovation curve flattens."
it "gemini fantasy"– The shift toward a highly regulated, product-centric culture – threatens the agility that allowed Kween to overtake Meta’s llamas in derivative model creation. For the global AI community, the loss of Junyang Lin is symbolic.
He was the primary bridge between China’s deep engineering talent and the Western open-source ecosystem. Without their advocacy, there are fears the project will lag "walled garden" Its strategy is similar to that of its Western rivals.
‘l‘Ewing was not your choice’
The technical brilliance of the Qwen3.5 release has been overshadowed by the heartbreak of its creators. On social media, there is a sense of mourning rather than celebration among members of the model-making team:
Quen contributor Chen Cheng clearly alluded to the forced departure, writing in a post on X: "I am really very sad. I know it wasn’t your choice to leave… I honestly can’t imagine Quen without you."
Lee suggested that the exit signals the end of broader ambitions, such as a planned Singapore-based research centre: "Thanks to Junyang, Kwen could have a Singapore base. But now that he is gone, there is no reason left to stay here."
What will happen to Quen’s open source AI efforts from now on?
The known facts are simple: Quen has never been technologically strong, yet its founding core has been destroyed. As Alibaba prepares to face investors for its fiscal Q3 earnings report on March 5, the story will likely focus on "Capacity" And "commercial scale."
For enterprises currently excited by the 60% cost reduction promise made by Qwen3.5, the immediate future is bright.
But for the larger AI community, the price of that efficiency may be the loss of what was formerly the most vibrant open-source lab.
As Hao Zhou takes the reins, the world is watching to see if Quan persists "model for the world" Or simply become a component in Alibaba’s corporate bottom line.
<a href