Qwen3.8-Flash, a new-generation MoE multimodal model from Qwen. It supports text, image and video input, with a 1M-token context window. It delivers fast inference and excels at coding, agent workflows and visual comprehension.