【#Tech24H】In OpenRouter’s latest weekly global AI large-model call-volume ranking (October 19 to 25), seven of the top ten global large models are from China. The top five were DeepSeek V4.1-Flash-1017, DeepSeek V4.1 Flash, MiniMax M3.1 Flash (free), GLM5.3 Flash, and Tencent Hy4 preview. The top ten show that lightweight Flash versions have become the absolute traffic drivers of platform traffic, and heavy-parameter flagship models are no longer developers’ first choice. In addition, million-token long-context windows have become almost standard for top new models, with long documents, video understanding, and continuous agent tasks becoming essential needs. Moreover, large-model iteration cycles keep shrinking, new versions are launched densely, traffic switching is becoming faster, and the advantage window for a popular model is constantly shortening. Overall, last week’s total global AI large-model call volume was 188 trillion tokens, up 14.63% week-on-week. Among this, the weekly call volume of Chinese AI large models on the list was 93.54 trillion tokens, up 18.14% week-on-week.[ By Zhang Liyan | Tang Ruohan ]