China overtakes U.S. in global open-source AI model market: study

CGTN – A new study shows China has overtaken the United States in the global market for open-source artificial intelligence (AI) models, gaining a significant advantage in one of the world’s most technology-driven fields.

The study, by the Massachusetts Institute of Technology (MIT) and open-source AI startup Hugging Face, reported that Chinese-developed open-source AI models made up 17.1 percent of global downloads over the past year, surpassing the U.S. share of 15.8 percent for the first time. Most of the downloads for Chinese models come from DeepSeek and Alibaba’s Qwen.

Open-source models, which developers can freely download, modify, and integrate, are accelerating product development for startups and advancing AI research forward worldwide. Their rapid adoption is shaping the future of the technology.

“Open source has become a powerful engine for global IT development, and in the era of AI driven by large models, its momentum is remarkable,” said Ni Guangnan, an academician of the Chinese Academy of Engineering, at the 2025 OpenAtom Developers Conference in Beijing on November 21.

“China is now the world’s largest provider of open-source large models. Models such as Qwen, DeepSeek, and Kimi rank among the top on evaluation platform LMArena,” he said.

China’s 15th Five-Year Plan (2026-2030) calls for a higher level of opening up and mutually beneficial cooperation, stating that China’s push to promote and develop open source aligns with global trends.

Chinese-led open-source communities are also expanding rapidly overseas. A U.S. report cited by Ni found that 80 percent of American AI startups use Chinese open-source models. He said China’s open-source AI ecosystem embraces inclusiveness and global access, pooling international developer talent, facilitating technological exchange, and injecting fresh momentum into innovation.