Chinese AI Models Gain Global Edge as Usage Surges Past US Competitors
Chinese AI Models Outperform US Counterparts in Weekly Usage
In a notable development for the global AI industry, Chinese large language models have surpassed their American counterparts in weekly usage metrics for two consecutive weeks. The latest data from OpenRouter reveals Chinese models processed 46.9 trillion tokens during March 9-15 - an 11.83% increase from the previous week - while US model usage declined by 9.33% to 32.94 trillion tokens.
The Leading Contenders
Three Chinese models currently dominate the global rankings:
- MiniMax M2.5 maintains its five-week streak at the top with 17.5 trillion tokens processed
- Stephen Star Step3.5Flash surged 79% to claim second place with its rapid response capabilities
- DeepSeek V3.2 holds strong in third position at 10.4 trillion tokens
The real surprise comes from newcomer Hunter Alpha, which debuted on March 11 and immediately captured seventh place globally with 6.66 trillion tokens processed.
Hunter Alpha: The Rising Star
This mysterious new entrant specializes in intelligent agent applications, featuring:
- Trillion-parameter architecture
- Ultra-long context capacity of 1 million tokens
- Exceptional performance in complex reasoning and multi-step tasks
The model shows particular promise when integrated with frameworks like OpenClaw, demonstrating remarkable reliability and instruction execution accuracy.
Key Points:
- Chinese AI model usage grew nearly 12% week-over-week
- US model usage declined by over 9% during same period
- MiniMax maintains five-week dominance at top of rankings
- Hunter Alpha achieves remarkable debut despite late entry
