Alibaba says its Qwen family of open-weight models has passed 3 billion downloads, putting the Chinese lab ahead of Meta and Google in the global open-weight race. The company made the claim in an emailed statement, and it is the clearest signal yet that Alibaba is winning on raw reach among open model families, even if the underlying platform reports a smaller number.
What the 3 billion figure rests on
The headline number comes from Alibaba directly, not from Hugging Face, the platform usually used to measure open-model activity. The Hugging Face State of Open Models report, published a day before Alibaba’s announcement, counted 2,045 million Qwen downloads for the year, or 2,061 million when every related repository is included. The gap between the two figures is roughly a third, which means the 3 billion claim is larger than anything the cited hub actually recorded.
Derivatives: 300,000 versus 151,448
The same pattern shows up in derivative models, the fine-tuned and rebuilt versions other developers publish on top of a base model. Alibaba says its ecosystem has produced more than 300,000 Qwen-based derivatives. Hugging Face counts 151,448 Qwen derivatives on its Hub, which is still 2.6 times the size of Meta’s entire derivative footprint on the same platform. Either way, the Qwen family leads by a wide margin.
Why the gap between the two numbers exists
Hugging Face is explicit about what its download counts actually measure. Downloads count activity inside its own hub only, and do not capture API usage, private deployments, or models distributed through other channels. Alibaba runs its own distribution platform in ModelScope, and that traffic is not part of the Hugging Face tally. So the 2,045 million Hugging Face figure and the 3 billion Alibaba figure are not measuring exactly the same thing, even though both are being used to describe Qwen’s reach.
What the report actually says about Qwen’s role
Beyond the raw numbers, the State of Open Models report describes Qwen as part of the default workflow for developers deciding which models to fine-tune and deploy. That phrasing, included in the report itself, treats Qwen as a baseline choice rather than a single product, which is the kind of position open-weight families are trying to build.
How Alibaba got here: scale and distribution
Alibaba has open-sourced more than 460 models under permissive licences and pushes them through its own cloud into Southeast Asia and Africa. That wide net of distribution is what produces the derivative counts Hugging Face sees on its platform. The company has also said it wants to start charging the heaviest users of those same open models, which points to a longer-term plan to convert open reach into paid usage.
Who is answering, and from where
American labs are pushing back. Meta and Nvidia have both shipped new open models in recent weeks, and Nvidia has said publicly that it is trying to catch up with Chinese labs on open weights. Meta’s Llama family remains the most direct competitor to Qwen by derivative count, with Hugging Face placing it at roughly one-third the size of Qwen’s footprint on the Hub.
The bigger risk to Qwen’s lead may be domestic
The threat to Qwen’s open-weight run may not come from the United States at all. Beijing has been weighing curbs on overseas access to the country’s strongest models, which would directly reduce the global reach that produced these download numbers in the first place. Any limit on international distribution would shrink both the 3 billion Alibaba claim and the 2,045 million Hugging Face count at the same time.
FAQ
How many downloads does Qwen have?
Alibaba says its Qwen family has passed 3 billion downloads in total. The Hugging Face State of Open Models report, published a day earlier, counted 2,045 million Qwen downloads for the year, or 2,061 million including every related repository.
How many derivative models are built on Qwen?
Alibaba says its ecosystem has produced more than 300,000 Qwen-based derivatives. Hugging Face counts 151,448 Qwen derivatives on its Hub, which is 2.6 times Meta’s total derivative count on the same platform.
Is Qwen ahead of Meta and Google in open-weight downloads?
Yes. Alibaba’s claim puts Qwen ahead of Meta’s Llama family and Google’s open-weight offerings on downloads, and Hugging Face’s derivative count places Qwen 2.6 times ahead of Meta’s footprint on its Hub.
Related coverage
- Cisco Antares: Open-Weight AI Models for Vulnerability Localization
- Poolside Laguna S 2.1 open-weight coding model
Want help with this in Myrtle Beach or Long Island? Talk to us.
NewSunSEO
Grow in Google and AI Search
AI-SEO, local search, paid ads and websites from one team, with offices on Long Island and in Myrtle Beach.
- AI-SEO
- Local AI-SEO
- AI Consulting
- PPC
- Web Design
- Web Development
- Long Island Office(631) 320-1700
- Myrtle Beach Office(843) 492-7131
Request a Free Proposal
Tell us about your business and where you want to grow.
We will reply by email to talk through your goals and what a proposal would include.
We use your details only to reply to you and send what you asked for. See our Privacy Policy.
