**[Bloomberg: Performance Gap Between Top AI Models in China and the U.S. Narrows to 3%, Hits Historic Low]** Beating AI News Flash: Bloomberg Industry Research (BI) reports that as Chinese AI labs like DeepSeek make progress, the performance lead of U.S. AI companies over their Chinese counterparts has dropped to a historic low. Following the release of DeepSeek V4.1 Flash in September, China's top models lagged behind their U.S. counterparts by only 3% in benchmark tests, a significant improvement compared to approximately 9% in May and 15% at the beginning of the year. Last month, DeepSeek V4.1 Flash ranked sixth globally on LiveBench with a score of 81.1, close to Anthropic's highest score of 83.4. BI senior analyst Robert Lea believes the model has achieved performance levels comparable to leading systems from Anthropic and OpenAI. However, among the top 15 models on LiveBench, only three are from China. The progress of China's AI industry stems from enhanced technical expertise and the ability to optimize models for domestic hardware, which also raises questions about the effectiveness of U.S. chip export restrictions. Nonetheless, China's AI sector still faces commercialization pressures. Lea predicts that low-profit token supplies, intense price wars, and competition among more than 1,100 large language models may prevent the industry from achieving profitability by 2030. Currently, ByteDance's Doubao leads in monetizing AI applications, while DeepSeek and Tencent's chatbots remain free to use. [Original Link]