Deepseek API大降价，开发者成本可降超50%

Core Viewpoint - DeepSeek has released its new model DeepSeek-V3.2-Exp, which significantly reduces service costs and enhances model inference efficiency through a sparse Attention architecture [1][5]. Group 1: Model Updates - The DeepSeek-V3.2-Exp model was officially launched on September 29, with updates across its app, web, and mini-program platforms [1]. - The previous version, DeepSeek-V3.1, was updated to DeepSeek-V3.1-Terminus on September 22, showing a performance improvement of over 36% [5]. - The API pricing for DeepSeek has been reduced by over 50% due to the new model's cost efficiency [1]. Group 2: Market Impact - Following the release of DeepSeek-V3.1, there was a notable surge in stock prices for several domestic chip industry companies [5]. - Analysts highlight that the rapid updates and bug fixes in open-source models are crucial for retaining developers in a competitive landscape [5]. Group 3: Industry Insights - The AI computing demand is expected to grow due to large model training and inference, benefiting various segments of the computing industry, including AI chips and servers [6]. - DeepSeek's innovations and performance position it competitively against global leaders, reshaping investor perceptions of Chinese tech companies' capabilities [5].