Chinese Open-Source AI Models Surpass US, Capturing 41% of Global Downloads and Dominating Top Six Rankings

As the global artificial intelligence race intensifies, the open-source model battlefield is rapidly reshaping the industry landscape beyond just computational power battles between closed large models. The latest data from Hugging Face, the world's largest open-source AI model platform, shows that Chinese open-source models now account for 41% of monthly and total downloads on the platform, surpassing the United States for the first time and becoming the dominant force in the global open-source ecosystem. This data was recently cited by Hugging Face CEO Clem Delangue, sparking widespread attention across the tech community and social media.

Hugging Face, a hub for developers and researchers worldwide, currently hosts nearly 3 million public models and 1 million public datasets, with a new code repository created on average every seven seconds. As the open-source ecosystem flourishes, more than half of the Fortune 500 companies have begun deploying private or open-source models on the platform.

Notably, the rise of Chinese open-source power extends beyond download figures. This trend is equally evident on OpenRouter, an AI model calling platform.

Currently, the top six most popular models on OpenRouter are all developed by Chinese institutions, including Tencent, Xiaomi, DeepSeek, MiniMax, and Zhipu AI.

In contrast, Anthropic from the United States, previously a long-term leader, now sees its Claude Opus 4.7 model ranked seventh, indicating strong market recognition and adoption of Chinese open-source models.

The rise of Chinese open-source models is no accident, but built on rapid technical iteration. For example, Zhipu AI’s recently released open-source model GLM-5.2, designed specifically for agentic coding tasks, performs at the same competitive level as Anthropic’s latest closed-source models in multiple benchmark tests.

As DeepSeek and other Chinese AI labs continue to release powerful open-source models, these tools are no longer mere "affordable alternatives," but possess the robust capabilities to compete with top US models in complex scenarios. Infrastructure data also confirms this: according to Vercel platform statistics, as of June 2026, open-source models have handled nearly one-third of the platform’s AI requests, taking on a large volume of infrastructure-intensive workloads, while traditional closed-source models are gradually retreating to high-cost premium domains.

The core logic driving global enterprises toward open-source models lies in the pursuit of "control" and "cost efficiency." As leading US AI labs shift their enterprise services to token-based usage pricing, expensive API fees have become a heavy burden for businesses. Delangue has analyzed that tech companies and developers are realizing core capabilities should not be outsourced to uncontrollable "black-box APIs."

In contrast, open-source models allow enterprises to host and fine-tune them on their own infrastructure, drastically reducing maintenance costs while ensuring data security and autonomy. Many well-known tech companies in Europe and the US, including software developers, engineering consulting groups, and logistics platforms, have begun assigning low-level or highly repetitive AI tasks to these cost-effective Chinese open-source models.

Open-source acts as an accelerator, redefining AI leadership. Delangue holds an optimistic and open view of the future of AI development, believing open-source is the key catalyst accelerating global AI advancement. If China can maintain its leading position in the open-source ecosystem, it could gain a significant advantage in the global AI leadership race within the next one to two years.

Although some industry leaders, such as Anthropic CEO Dario Amodei, express concerns about the risks of releasing overly powerful models, the mainstream open-source community believes that locking core innovation behind the doors of a few institutions is not a sustainable path for technological democratization.

The current reality is clear: Chinese open-source models have successfully transformed from "followers" to "definers." In the future AI ecosystem, regardless of how US proprietary models evolve, the open-source infrastructure supplied by China has become an indispensable foundation for global technological development.