Qwen3.8-27B is a 27-billion-parameter dense vision-language model for text, image, and video understanding, coding, multimodal reasoning, long-context processing, tool calling, and agentic tasks.