Chinese AI models surpass US-made counterparts in usage

Chinese AI models are now being utilized more frequently than their US-made counterparts, driven by heightened adoption across various applications such as document processing and customer interaction in China. Recent data shows that calls to Chinese AI models have surpassed those made to US models, as these Chinese systems are becoming increasingly competitive, now only months behind US technologies in terms of capabilities. Furthermore, their lower cost and greater openness have facilitated local experimentation and deployment, contributing to their rising share in global AI usage.

Chinese AI models: Chinese AI models are large-scale artificial intelligence systems developed by China-based companies and research labs, including popular families like Qwen, DeepSeek, Kimi, and MiniMax. They are central to this news because recent usage data shows these models collectively being called more frequently than US-made counterparts, reflecting China’s rapid AI adoption and the growing global competitiveness of its models.
US-made AI models: US-made AI models are frontier large language and multimodal systems created by American firms such as OpenAI, Anthropic, Google, and others, which have historically led in benchmark performance and global adoption. They feature in this news as the reference group being overtaken in total usage by Chinese models, highlighting a shifting balance in practical, real-world deployment even as US models still retain an edge on the most difficult benchmark tasks.

Adoption: Recent platform data indicates that calls to Chinese AI models now exceed those to US-made models, driven by intensive use across document processing, software generation, customer interaction, and operational workflows in China.
Capabilities: Analyses of newly released Chinese models find that they are now only months behind US frontier systems in capability, competing closely on software engineering, agentic tasks, and coding benchmarks.
Cost_and_Openness: Commentaries highlight that leading Chinese models are generally cheaper and more open-weight than many US models, encouraging experimentation and local deployment and helping them capture a growing share of global AI usage.