Alibaba Releases Qwen3.8-Omni-Flash, Omni-Modal Model With Agent Capabilities

Alibaba's Qwen team unveiled Qwen3.8-Omni-Flash on September 18, a native omni-modal model built from the ground up to handle audio, video, image, and text understanding alongside autonomous agent capabilities like tool use and task planning. The model natively processes understanding and reasoning across all modalities within a single forward pass, distinguishing it from systems that stitch together specialist components.