Billions of downloads and leadership in open-source: how Qwen from Alibaba reshaped the AI market
Over the past six months, the ecosystem of open AI models Qwen from Chinese company Alibaba has reached an impressive milestone—over 3 billion downloads. This is not just a number, but a marker of a tectonic shift in an industry where open weights are becoming the main battleground for developers' minds.
The corporation has made over 460 neural networks publicly available, based on which the global community has created about 300,000 derivative models. Such a scale indicates that Qwen has become the de facto standard for fine-tuning and customization.
Hugging Face Analytics: Numbers Don't Lie
Independent statistics from the Hugging Face platform paint an even more telling picture. In just seven months of this year, Qwen models were downloaded 2.05 billion times, and the number of derivative repositories reached 151,448. For comparison, Google's figure is only 82,506. Meanwhile, Google's total downloads for 2026 are estimated at 418 million, and Meta's at 227 million. It is important to understand that these data reflect activity only within the Hugging Face ecosystem, excluding API requests and private deployments, which makes the real gap even more significant.
The dynamics are especially noteworthy: the number of Qwen derivative repositories grows by 180–210 units daily. Of the 28,531 GGUF conversions (a format for local execution), only 54 were created by Alibaba itself—the rest is generated by the community. This testifies to incredible virality and developer trust.
The Secret to Success: From Sub-Billion to Trillions
My attention is drawn to Alibaba's strategy, which differs radically from Western labs. Qwen's success rests on three pillars: regular updates, coverage of all scales—from compact versions to the giant Qwen3.8-Max with 2.4 trillion parameters—and the Apache 2.0 license, which does not restrict commercial use.
The breadth of the model lineup became the key factor. Data show that 83% of all downloads in the platform's history come from models with fewer than 1 billion parameters, while giants over 100 billion account for only 1%. This is precisely why labs focused on huge LLMs are falling behind. For example, Moonshot AI, which almost never releases models smaller than 70 billion, has gathered only 37 million downloads—roughly 55 times fewer than Qwen.
The Geopolitics of Open Source
The analysis also reveals fundamental differences in the approaches of the US and China. In China, 59% of models with more than 20 billion parameters are released under the Apache 2.0 license, and none of them have restrictions on commercial use. In the US, the picture is the opposite: 41% of such models are distributed under proprietary terms, and 30% have no specified license at all. This is a strategic advantage for China that is hard to overstate.
It is also telling that in the US, the leaders in the number of new open models are not AI labs but chip manufacturers—AMD and Nvidia, each releasing over 200 repositories. This indicates a shift in focus from fundamental research to infrastructure.
My conclusion: We are witnessing not just competition among models, but a battle of ecosystems. Alibaba is winning through total coverage and openness, while the West tries to maintain leadership through control over computing power. In the coming years, it will be the accessibility and flexibility of open-source solutions, not the number of parameters, that determine real influence in the AI industry.