Chinese AI Models Outpace U.S. Rivals in Global Token Usage
Chinese AI models have surpassed U.S. alternatives in global weekly token consumption for five consecutive weeks, driven by aggressive pricing and industrial scaling.
Chinese artificial intelligence models have outperformed United States models in global weekly token consumption for five consecutive weeks. Data from OpenRouter shows that from March 30 to April 6, global usage of Chinese models reached 12.96 trillion tokens, compared to 3.03 trillion for U.S. models. All six of the world's most-used AI models are currently Chinese, including two from Alibaba Group's Qwen and the Doubao model from Beijing Volcano Engine Technology Co., Ltd., which processes over 120 trillion tokens daily.
MiniMax and other developers have achieved a competitive edge by offering performance comparable to U.S. alternatives at approximately one-tenth of the cost. This growth is attributed to algorithmic innovations that reduce token consumption and China's comprehensive industrial chain and power supply. By December 2025, generative AI users in China reached 602 million, accelerating the integration of AI into industrial design and office collaboration.
To sustain this momentum, the Ministry of Industry and Information Technology is piloting computing banks and computing supermarkets. These initiatives aim to financialize computing power as a tradeable resource for small and medium-sized enterprises. China intends to establish a widely accessible, low-cost computing service system across major industry categories by the end of 2028.