The latest industry data shows that last week, the total global call volume of AI large models reached 188 trillion Tokens, up 14.63% from the previous week. Among them, Chinese AI large models recorded a weekly call volume of 93.54 trillion Tokens, up 18.14% from the previous week, surpassing the United States for twenty-six consecutive weeks and firmly holding the global top spot; during the same period, the weekly call volume of US AI large models was 22.47 trillion Tokens, up 29.44% from the previous week.
In terms of specific model performance, the global top five in weekly call volume last week were all Chinese AI large models. Among them, the newly launched DeepSeek V4.1-Flash-1017 powerfully claimed the global top spot with a weekly call volume of 33.6 trillion Tokens and a quarter-over-quarter increase of up to 801%. The remaining spots in the top five were occupied by DeepSeek V4.1Flash, the newly listed MiniMax M3.1Flash (free), Zhipu GLM5.3Flash, and Tencent Hy4preview.
In addition, elsewhere on the list, Xiaomi MiMo-V2.6-Flash dropped to eighth place, while Zhipu GLM5.3, which ranked eighth the previous week, unfortunately fell off the list. The newly listed models showed varying degrees of optimization and improvement in performance and cost-effectiveness, continuing to drive the adoption of AI large models in real-world applications.