Alibaba's Qwen 3.6 Max reportedly outperforms Claude 4.5 Opus on instruction-following and agentic coding

Alibaba released Qwen 3.6 Max, which independent testers say outperforms Claude 4.5 Opus and Zhipu's GLM 5.1 on instruction following, agentic coding and multimodal processing, with sharper visual reasoning and document analysis. The release lands alongside Moonshot AI's $2B raise at a $20B valuation as the Chinese open-weights ecosystem closes the gap with US frontier labs.