Qwen open-sources Qwen3.8-Flash-Next, a preview of the Qwen4 architecture
Alibaba's Qwen team open-sourced Qwen3.8-Flash-Next, a multimodal MoE that previews the architecture that will underpin Qwen4. It packs a 125B-parameter main model plus 51B N-gram embeddings, yet only 6B parameters activate per token. Training it costs roughly a ninth of Qwen3.7-Plus while beating it on coding and office tasks, and the new hybrid attention design is what the full Qwen4 family will build on.