Alibaba confirms Qwen 4 is in training on a new architecture

At the Apsara Conference in Hangzhou, Alibaba confirmed that Qwen 4 is in training on a "new-generation architecture" but gave no date, specs or licensing terms. The current flagship Qwen3.8-Max is a ~2.4T-parameter MoE with ~95B active parameters and a 1M-token context; Qwen3.8-Flash-Next is described as a preview of the Qwen 4 architecture.

• Qwen 4 is in training on a new architecture; no date or license disclosed
• Qwen3.8-Max: ~2.4T-parameter MoE, ~95B active, 1M-token context
• Qwen3.8-Flash-Next shipped in late August as a Qwen 4 architecture preview
• Manifold market prices a launch before 1 November 2026 at roughly 74%