Qwen 35B removed from commits, signaling possible cancellation
Qwen's 35B MoE model vanished from latest commits, sparking community backlash.
Deep Dive
The article reports that a recent commit removed the 35B model, which the author sees as confirmation it won’t be released. The author argues that the 35B MoE is widely used and that users need to make noise on X, Hugging Face, and other online spaces — because if the team doesn’t know how popular it is, there’s no reason to release it for a group that doesn’t speak up.
Key Points
- Qwen 35B removed from recent commits on the official repository, indicating it may never ship
- The model is suspected to be a Mixture-of-Experts (MoE) design with significant adoption for local deployment
- Community backlash is growing on X and Hugging Face, demanding an official response from Alibaba's Qwen team
Why It Matters
If Alibaba cancels the 35B, developers lose a popular efficient MoE option, shaking confidence in Qwen's open-source roadmap.