Alibaba's Qwen family โ a full open-weight stack from 0.5B to 110B+ with vision, coding, and agent modes.
Qwen (้ไนๅ้ฎ) is Alibaba's flagship model family and one of the strongest open-weight contenders globally. The Qwen Chat product bundles chat, vision, code, and long-context modes in one interface, while the underlying models โ from a 0.5B edge model to 110B+ MoE โ are downloadable and commercially usable under permissive licenses. For teams in APAC or anyone building multilingual products, Qwen is the regional default.
Who it's for: Multilingual teams, APAC businesses, and engineers who want a full open model family with vision and agent support.
Strong in Chinese, English, and 20+ languages out of the box.
Qwen-VL for images and Qwen-Coder for repositories.
From 0.5B edge models to 110B+ MoE โ one family, every use case.
Download and ship in production under permissive terms.
For multilingual and open-weight needs, Qwen is the most complete family after DeepSeek. Use it when you need vision, code, and chat from one downloadable stack.