According to a report by Bloomberg Intelligence, the performance gap between top-tier AI models from China and the United States has narrowed to 3%. This shift follows the release of the DeepSeek V4.1 Flash model in September, marking a significant decline from the 9% gap recorded in May and the 15% difference observed earlier this year.
The rise of Chinese AI capabilities is attributed to deeper technical expertise and the ability of researchers to optimize models for domestic hardware. Analyst Robert Lea noted that these advancements challenge the long-term sustainability of US technological dominance in the AI sector and raise questions regarding the effectiveness of existing US export restrictions.
In the latest LiveBench global rankings, DeepSeek V4.1 Flash secured the sixth position with a score of 81.1 points. This performance places the startup's model in direct competition with leading systems from Anthropic, which currently holds the top score of 83.4 points. Despite this progress, only three of the top 15 models listed on LiveBench originate from China.
While the performance metrics show rapid convergence, experts caution that rankings remain fluid. Furthermore, Robert Lea emphasized that achieving high performance in benchmarks does not automatically guarantee commercial success, noting that monetization remains a significant challenge for AI companies regardless of their current standing in global leaderboards.
