Qwen 3.5 Small

The small tier of Qwen 3.5 is Alibaba's set of compact, natively multimodal models built for on-device and latency-sensitive use. Ranging from 0.8B to 9B parameters, they understand both text and images out of the box.

Each model supports switchable thinking and non-thinking modes, letting you turn step-by-step reasoning on for hard problems or off for fast replies, with language coverage spanning 201 languages.