On July 22, Alibaba Cloud's Tongyi Qianwen large model, Qwen, officially unveiled an updated version of its flagship Qwen3 model. The new iteration, designated as Qwen3-235B-A22B-Instruct-2507-FP8, represents a significant upgrade from the previous Non-thinking mode-based Qwen3-235B-A22B-FP8. This enhancement spans various crucial domains including instruction adherence, logical reasoning, text comprehension, mathematics, science, programming, and tool utilization, marking a substantial improvement in the model's overall capabilities. The updated Qwen3 model is now available as open-source on ModelScope and HuggingFace, providing broader access to its advanced functionalities.