Imagine shipping a product so good that even your biggest competitors have to tip their hats to it, only to pack up your desk and leave the company the very next day. That is the confusing and somewhat alarming scenario currently unfolding at Alibaba.
The Chinese tech giant’s AI research division has been on a tear lately. Their Qwen (Tongyi Qianwen) models have become global darlings of the open-source community, recently challenging Meta’s Llama series in download metrics. But just as the team celebrated their latest victory—the release of the Qwen3.5 small model series on March 3, 2026—the celebration turned into a wake.
Within 24 hours of the launch, the technical lead and public face of the project, Junyang "Justin" Lin, announced he was out. And he wasn’t the only one. This sudden shakeup has sent shockwaves through the machine learning world, leaving many of us asking: What exactly is going on inside Alibaba?
What happened to the Qwen leadership team?
The timeline here is startlingly compressed. On March 3, the team dropped the Qwen3.5 series. By the next day, the team’s structure seemed to implode.
Justin Lin took to X (formerly Twitter) with a brief, emotional message: "me stepping down. bye my beloved qwen." It was a stark departure for a figure who has been instrumental in steering Qwen from a small lab project into a global heavyweight.
He wasn’t alone. Researcher Kaixin Li also announced their departure. But the most telling detail came from fellow contributor Chen Chang, who replied to the news with a comment that suggests this wasn’t just a case of burnout or moving on to greener pastures. Chang wrote, "I’m truly heartbroken. I know leaving wasn’t your choice. Just last night, we were side by side launching the Qwen3.5 small model."
That phrase—"wasn’t your choice"—is doing a lot of heavy lifting. It hints that despite the external success, internal friction at Alibaba has reached a breaking point.
How good are the new Qwen3.5 models?
The irony of these departures is that they coincide with a technical triumph. The Qwen3.5 Small Model Series isn’t just another incremental update. Ranging from 0.8B to 9B parameters, these models feature a new "Gated DeltaNet" hybrid attention mechanism.
The performance is turning heads. Elon Musk, who is usually focused on his own xAI initiatives, publicly praised the release, calling out the models for their "impressive intelligence density." When you have the CEO of Tesla and xAI complimenting your efficiency immediately after launch, you know you’ve built something significant.
These models were designed to punch above their weight class, offering high performance with lower compute requirements—exactly the kind of innovation the open-source community craves. Yet, the architects behind this success are no longer at the helm.
Is Alibaba shifting its AI strategy?
If the team is successful, why force them out? The answer likely lies in a broader strategic pivot occurring within Alibaba Group. Reports indicate that the company is undergoing a significant internal restructuring. They are unifying their various AI initiatives under the single "Qwen" brand, but the end goal appears to be changing.
Alibaba seems to be moving away from pure, research-driven exploration and toward aggressive commercialization. There are reports that the company is preparing to launch AI-enabled hardware, specifically smart glasses, later in 2026. This follows the January release of Qwen 3-Max-Thinking, which claimed to outperform competitors like DeepSeek V4.
It’s a classic tension in the tech world: the researchers want to publish papers and open-source code, while the executives want to ship consumer products and lock down IP. The departures suggest that the executives have won this round.
Will this impact the open source community?
This is the biggest fear for developers who have come to rely on Qwen. The team has been incredibly prolific, shipping dozens of generalized and specialized models since last summer, most of them open source and free. They established themselves as a primary challenger to Western models, offering a vital alternative in the ecosystem.
The forced exit of key technical leadership is being viewed by some as a "kneecapping" of the team. If the visionaries who championed open weights are gone, replaced by product managers focused on selling smart glasses, the pipeline of free, high-quality models could dry up. This could drive talent toward rival labs or new startups, potentially altering the competitive landscape just as Qwen was taking the lead.
What This Really Means
This isn’t just corporate drama; it’s a signal that the "Golden Era" of funded, altruistic open-source research at big tech firms is tightening. Alibaba has likely calculated that they have squeezed enough prestige and brand awareness out of the open-source community and are now pivoting to Return on Investment (ROI). The losers here are the developers and researchers who relied on Qwen as a high-performance, open alternative to locked-down US models. While the brand name remains, the ethos that drove Qwen’s rapid ascent left the building along with Justin Lin.
![Illustration related to Why Justin Lin Left Alibaba: Qwen Team Split [Analysis]](https://bytewire.press/wp-content/uploads/bytewire-images/2026/03/alibaba-qwen-team-departures-justin-lin-open-source-235abdd794.webp)
![Diagram related to Why Justin Lin Left Alibaba: Qwen Team Split [Analysis]](https://bytewire.press/wp-content/uploads/bytewire-images/2026/03/alibaba-qwen-team-departures-justin-lin-open-source-db85694426.webp)
