According to a 14 August report by the Hugging Face platform, Qwen passed three billion downloads in six months. Google's open models reached 418 million and Meta's 227 million.
The report by developer platform Hugging Face records more than three billion downloads of Qwen models over six months. By comparison, Google's open models reached 418 million in the same period and Meta's 227 million. The Qwen team says more than 460 models have been open-sourced, spawning over 300,000 derivative models.
The report contains a further finding: in almost every month of 2026 the largest open model from a Chinese lab was bigger than anything released by an American lab. Chinese monthly ceilings ranged from 754 billion to 2.78 trillion parameters, while the American ceiling stayed below 130 billion in five of seven months.
For European users this is not a question of national preference but of supply chain. If a growing share of openly available base models originates in China, that affects provenance checks, licence terms and security requirements in every project built on open weights.
What this means for decision-makers
- Record which open models are actually used in your applications and in those of your service providers.
- Review open model licence terms before production use; they differ substantially and they change.
- Define which data classes may run on a self-hosted open model and which may not.
This story was produced automatically from the source named above and checked by software before publication. The image is symbolic and shows neither the event nor a real person. How this paper is made