en.Wedoany.com Reported - Recent global AI large model call volume data shows that China's weekly call volume reached 18.81 trillion tokens, ranking first globally for eight consecutive weeks. Monitoring by the OpenRouter platform indicates that from June 15 to 21, the global total call volume was 46.7 trillion tokens, rising for nine consecutive weeks.
During that week, the US AI large model weekly call volume was 5.76 trillion tokens, a month-on-month increase of 0.70%; the Chinese market saw a month-on-month increase of 2.12%, marking four consecutive weeks of growth. Leading for eight consecutive weeks highlights China's core position in the global AI application deployment landscape, breaking the established pattern dominated by overseas models.
The domestic AI large model market exhibits characteristics of top-tier concentration and tiered differentiation. DeepSeek-V4-Flash has topped the domestic call volume rankings for five consecutive weeks, leveraging its performance and cost-effectiveness, with a weekly call volume of 4.94 trillion tokens, a month-on-month increase of 12%. Xiaomi MiMo-V2.5 rose to second place with a weekly call volume of 3.94 trillion tokens, a month-on-month increase of 10%. MiniMax M3 ranked third with 3.77 trillion tokens, a month-on-month decrease of 13%. Tencent Hy3preview fell to fourth place with a weekly call volume of 3.63 trillion tokens, a month-on-month decrease of 12%.
Zhang Yi, CEO of Guangzhou Aimei Data Gathering Information Consulting Co., Ltd., stated that China's lead in AI large models stems from two core advantages. First, the country's rich real-economy scenarios and massive national-level traffic entry points (such as WeChat, short videos, and smart government services) provide large models with vast amounts of real interaction data and a continuous foundation for iterative deployment. Second, domestic large models, relying on open-source ecosystems and extreme cost control, lower the entry barrier for enterprises, developers, and small and medium-sized merchants through low API call pricing, transforming AI technology into inclusive digital infrastructure.
Industry data indicates that the AI large model sector has entered a phase of high-quality development. As of June 22, there were 13,000 existing related enterprises in China. The total number of registrations in 2025 reached 395, with only 24 new additions since 2026, reflecting a trend toward rational growth. In terms of corporate capital structure, enterprises with registered capital of 50 million yuan or more account for 44.52%, those with 10 million to 50 million yuan account for 30.72%, together exceeding 75%; enterprises with registered capital below 1 million yuan account for only 3.21%. Regionally, first-tier cities account for 42.14% of related enterprises, while new first-tier cities account for 34.69%, with first- and second-tier cities together exceeding 75%.
Industry analysts state that the AI large model sector has shifted from explosive incremental expansion to stock optimization and quality improvement, with small and medium-sized enterprises lacking core technologies accelerating their exit. Zhang Yi pointed out that China's leading AI large model call volume and industrial clustering indicate the formation of a positive development cycle. In the future, with breakthroughs in core technologies and ecosystem improvements, domestic AI large models will progress from leading in application scale to leading comprehensively.
This article is compiled by Wedoany. All AI citations must indicate the source as "Wedoany". If there is any infringement or other issues, please notify us promptly, and we will modify or delete it accordingly. Email: news@wedoany.com
