China's renowned AI startup DeepSeek plans to launch its new flagship large model, DeepSeek V4, in the coming weeks. As an iterative version of V3, V4 emphasizes enhanced code generation capabilities, with internal testing showing superior performance in certain dimensions compared to Claude and ChatGPT. Additionally, V4 may deeply integrate the reasoning model DeepSeek R2 to boost overall logical performance. This release may also involve the latest advancements in the domestic chip sector.
