Deepseek API大降价，开发者成本可降超50%

Core Insights - DeepSeek has officially released and open-sourced the DeepSeek-V3.2-Exp model on the Hugging Face platform, with updates across its official app, web, and mini-programs [1] - The new model incorporates a sparse Attention architecture, which reduces computational resource consumption and enhances inference efficiency [1] - The introduction of the new model has led to a significant reduction in service costs, with API prices dropping by over 50% [1] Company Developments - On September 22, DeepSeek updated the DeepSeek-V3.1 model to the DeepSeek-V3.1-Terminus version, achieving a maximum single-item performance improvement of over 36% [3] - The API pricing for the previous version remained unchanged, with costs set at 0.5 yuan for 1 million tokens input and 12 yuan for output [3] - The rapid updates and enhancements in model performance are seen as crucial for retaining developers in a competitive open-source model landscape [3] Industry Impact - Analysts from Guosheng Securities highlight that DeepSeek's high performance, low cost, and innovative algorithms position it competitively against global standards, reshaping investor perceptions of Chinese tech companies' innovation capabilities [3] - The growth in AI model training and inference is expected to drive demand for AI computing power, benefiting various segments of the computing industry, including AI chips and servers [4]