Alibaba shifts Qwen AI strategy from open source to paid models

Saturday 11 April 2026 - 10:00

By: Dakir Madiha

Alibaba shifts Qwen AI strategy from open source to paid models

Alibaba is moving away from its open-source approach for Qwen, its flagship AI model family, and shifting toward paid, proprietary systems as it seeks stronger revenue growth. The change, first reported by the Financial Times, marks a major strategic pivot that could disrupt a large global developer community built around free access to Qwen models.

The shift became clear in early March when Junyang Lin, a central architect of Qwen’s open-source program, announced his resignation. Lin had led the technical direction of Qwen since 2022 and played a key role in establishing its rapid release cycle and open-access strategy.

His departure triggered an internal crisis meeting led by Eddie Wu at Alibaba’s Tongyi Lab. Several team members also resigned, reflecting tensions between leadership and engineers. Management had pushed to align AI development more closely with revenue goals, while the Qwen team operated with fewer resources than competitors such as ByteDance.

Less than two weeks later, Alibaba announced a major restructuring. The company created a new unit, the Alibaba Token Hub Business Group, consolidating its core AI teams, including Tongyi Lab and Qwen, under direct executive control. Wu said the group would focus on creating, distributing, and monetizing tokens, signaling a clear commercial direction.

Alibaba has already begun releasing closed-source models available only through APIs. Its latest model, Qwen3-Turbo, does not provide public access to model weights, breaking with the approach that helped Qwen become one of the most widely downloaded AI model families. Earlier proprietary releases, including Qwen2.5-Max and Qwen2.5-Plus, pointed to the same shift.

The stakes are high for developers. Qwen models have built a massive following, reaching nearly one billion cumulative downloads on Hugging Face. In February alone, downloads exceeded 150 million, more than double the combined total of several major competitors, including Meta and DeepSeek.

Despite strong adoption, Alibaba faces financial pressure. The company reported a 66 percent drop in net profit in its latest quarter, as heavy AI investment weighed on margins. Its cloud division continues to post triple-digit growth in AI-related revenue, but profitability remains uncertain.

The shift highlights a broader tension across the AI industry. Companies must balance open-source ecosystems that drive adoption with the need to generate returns on costly model development. Even firms like Meta are still struggling to show clear financial gains from large-scale open AI projects.