Chinese AI models dominate global token usage for 27 weeks running, OpenRouter data reveals

OpenRouter data released on November 2 indicates that global AI large model invocation volume reached 194 trillion tokens in the week spanning October 26 to November 1, marking a 3.19% week-over-week increase. During this period, Chinese large models demonstrated significant momentum, achieving a total weekly invocation volume of 99.55 trillion tokens, a 6.43% rise week over week. This volume significantly outpaced U.S. AI large models, which recorded 20.68 trillion tokens and saw a 7.97% decrease. This performance secured the 27th consecutive week where Chinese models led the global usage. In the top five global models by invocation volume, four were homegrown Chinese models. DeepSeek V4.1Flash1017 maintained the top spot, while MiniMax M3.1Flash experienced explosive growth, surging 690% week over week to claim eighth place in the rankings.