Chinese artificial intelligence models have achieved a historic milestone in global developer adoption, overtaking U.S.-developed systems in token usage for the first time according to OpenRouter data analyzed this week. The shift signals a significant rebalancing in the global AI landscape as China's tech sector accelerates innovation.
MiniMax's M2.5 model leads the pack with 1.7 trillion weekly tokens processed – nearly double the usage of second-place Google Gemini 3 Flash Preview. Three other Chinese systems (DeepSeek V3.2, Kimi K2.5, and GLM-5) collectively account for over 2 trillion additional tokens, demonstrating broad-based growth across multiple platforms.
National Business Daily statistics reveal the scale of this transition: Chinese models now process 5.16 trillion combined tokens compared to 2.7 trillion for U.S. counterparts. This 91% advantage comes as overseas developers increasingly integrate Chinese AI solutions into their workflows.
Industry analysts suggest the surge reflects both technical advancements and strategic partnerships. "The token metrics show Chinese models are being stress-tested at unprecedented scale," said a Shanghai-based AI researcher who requested anonymity. "Global developers appear confident enough to deploy them in production environments."
This development carries implications for business professionals monitoring Asia's tech sector and investors assessing AI market trends. The growing adoption of Chinese models could reshape global supply chains for AI-powered services while creating new opportunities in adjacent industries.
