Qwen: Qwen3.5-Flash

qwen/qwen3.5-flash-02-23

The Qwen3.5 native vision-language Flash models are built on a hybrid architecture that integrates a linear attention mechanism with a sparse mixture-of-experts model, achieving higher inference efficiency. Compared to the...