On March 2, 2026, Alibaba’s Qwen team shipped Qwen 3.5 Small Model Series: the 0.8B, 2B, 4B, and 9B. The benchmark numbers they shared are hard to ignore. The 9B variant is very close or even outperforms models that are 10 times its size across language, vision, and agentic benchmarks while running on a single consumer GPU.