Alibaba’s Qwen model family has crossed 3 billion downloads worldwide, the highest total for any open-weight series, according to a statement from the company. The figure puts the Chinese lab ahead of Meta and Google in the open-weights race and marks a turning point for how widely these models are being adopted outside of closed APIs.
What the downloads number actually counts
The 3 billion figure came from Alibaba directly. The Hugging Face report published the day before tells a slightly smaller story. The State of Open Models report from Hugging Face counted 2,045 million Qwen downloads over the same period, or 2,061 million when every repository on the platform is included. That puts Alibaba’s headline claim at roughly a third higher than the activity recorded on the platform most often cited as proof.
Hugging Face is clear about what its numbers do and do not cover. Downloads only reflect activity inside its own Hub. They do not count API usage, private deployments, or models distributed through other channels, and Alibaba operates its own distribution platform called ModelScope, which sits outside that count. The direction is not in question. The exact size of the lead depends on which platform you measure.
The derivative ecosystem is larger than the headline suggests
Alibaba says the Qwen ecosystem has produced more than 300,000 derivative models. Hugging Face counts 151,448 Qwen-based derivatives on its Hub, which is still 2.6 times the entire footprint Meta has on the platform. Both numbers point in the same direction: developers are fine-tuning and building on top of Qwen at a pace that outstrips the American competition on open weights.
A separate claim earlier in the summer that Qwen3 ranked as the world’s second-best model, with no public evidence offered to back it up, showed a pattern worth noting. Alibaba talks up its models in absolute terms. Outside platforms tend to show the same lead but with more conservative counts.
Why Chinese labs are pulling ahead on open weights
The method behind the lead is volume and reach. Alibaba has open-sourced more than 460 models under permissive licences and distributes them through its own cloud into Southeast Asia and Africa. The same report notes that Qwen has become part of the default workflow for developers deciding which models to fine-tune and deploy.
Alibaba has also said it wants to charge the heaviest users of those open models, a step that would change the economics of an ecosystem currently defined by free access. American labs are responding. Meta and Nvidia have shipped new open models in recent weeks, and Nvidia has stated openly that it is trying to close the gap with China on open weights.
What could slow the run
The bigger threat to Qwen’s momentum may be domestic. Beijing has been weighing curbs on overseas access to the country’s strongest models. If those restrictions land, they would cut off the international reach that produced these numbers in the first place.
For now, the position is clear. Alibaba’s Qwen family is the most downloaded open-weight series in the world on the platform that tracks such activity. Meta’s footprint is a fraction of the size. Google’s open-weight releases have not matched the same volume. The open-weights race, measured by downloads and derivatives, has a leader, and it is a Chinese lab.
FAQ
How many times have Alibaba’s Qwen models been downloaded?
Alibaba says Qwen models have passed 3 billion downloads. The Hugging Face State of Open Models report counted 2,045 million Qwen downloads this year, or 2,061 million across every repository. The gap reflects the difference between Alibaba’s own figure and activity recorded on Hugging Face’s Hub.
How many derivative models are built on Qwen?
Alibaba says the Qwen ecosystem has produced more than 300,000 derivative models. Hugging Face counts 151,448 Qwen-based derivatives on its Hub, which is 2.6 times the entire Meta footprint on the platform.
Which company leads the open-weight AI model race?
Alibaba’s Qwen family leads by downloads and derivatives, ahead of Meta and Google. American labs including Meta and Nvidia have released new open models recently, and Nvidia has said it is trying to catch up with China on open weights.
Related coverage
- Alibaba Qwen3.8-Max 2.4T Flagship, Open Weights Late
- Tech Firms Urge Washington Not to Restrict Open-Weight AI
- Poolside Laguna S 2.1 open-weight coding model released
This article summarizes reporting from thenextweb.com. See our editorial disclaimer for how our articles are produced.
Run a free scan to see your AI Visibility Score, SEO rating, and local citation accuracy.