China's open-source AI models account for nearly 30% of global usage, driving changes in the international AI landscape.

Foreign media: A report shows that China's open-source artificial intelligence models account for nearly 30% of global AI usage, and the global token usage volume of Chinese prompts ranks second, only after English.

OpenRouter, in collaboration with venture capital firm Andreessen Horowitz, based on empirical analysis of 100 trillion tokens, pointed out that Alibaba Cloud Tongyi Qianwen (Qwen), DeepSeek V3, and Moonshot K2 are the main forces driving the global adoption of Chinese large models.

The global share of China's open-source models rapidly increased from 1.2% at the end of 2024 to nearly 30% in 2025. The average weekly token share this year reached 13%, with significant growth in the second half of the year, approaching the 13.7% level of models from other countries around the world.

Although the United States has imposed restrictions on China's access to high-end GPUs from NVIDIA and AMD, China has formed a "quasi-equal" situation with the United States in AI large model development through models with international competitiveness. At the same time, Western closed-source models such as GPT-4o and GPT-5 still maintain a dominant position with about 70% of the market share.

Original article: toutiao.com/article/1850956374776832/

Statement: This article represents the personal views of the author.