Chinese open-source artificial intelligence models represent almost 30% of global usage, with Chinese-language prompts ranking second. Open-source large language model (LLM) usage surged this year due to Chinese-developed systems like Alibaba’s Qwen and DeepSeek’s V3. Western models like OpenAI’s GPT remain dominant with 70% global share.

According to OpenRouter’s study, Chinese open-source LLMs grew from 1.2% to nearly 30% of global share in a few months. Chinese LLMs matched the global average of 13.7% weekly token volume, showcasing China’s emergence as a major competitor in AI model development.

Despite US restrictions on Chinese firms, China has become a significant player in AI model development. Chinese open-source models demonstrate competitive quality, rapid iteration, and dense release cycles, challenging Western proprietary models’ dominance.

China’s open-source AI models by Alibaba Cloud and others have gained recognition for efficiency and low-cost adoption, making Chinese the second most used prompt language globally. China’s LLM tokens ranked fourth globally behind the US, Singapore, and Germany, leading to a fragmented landscape in AI model usage.

The global demand for open-source AI models has shifted from a DeepSeek-led monopoly to competition from Alibaba Cloud’s Qwen and Moonshot AI’s Kimi. The report highlights a widely distributed LLM usage landscape with no single model exceeding a 25% share.

Read more at Yahoo Finance: China’s open-source models make up 30% of global AI usage, led by Qwen and DeepSeek