Alibaba Record: Open-Source Qwen Models Surpass 3 Billion Downloads
Chinese tech giant Alibaba has reached an impressive milestone: the cumulative number of downloads of its open-source AI models from the Qwen family has exceeded 3 billion over the past six months. This figure demonstrates the colossal demand for neural networks from China and marks a significant shift in the global balance of power in the open artificial intelligence market.
To date, the corporation has released more than 460 models into the public domain, based on which third-party developers have created about 300,000 derivative solutions. This makes the Qwen ecosystem one of the largest and most dynamically developing in the industry.
Hugging Face Data: Dominance in the Ecosystem
An analysis by the Hugging Face platform, published ahead of Alibaba's announcement, sheds light on the nature of this success. Over seven months of the current year, Qwen models were downloaded 2.05 billion times, and the number of their derivative repositories reached 151,448. For comparison, the closest competitor from the US — Google — shows significantly more modest results: 82,506 derivatives and 418 million downloads over the same period. Meta's models lag even further behind with 227 million downloads.
It is important to emphasize that Hugging Face statistics only account for activity within its own ecosystem, excluding API requests and private corporate deployments. Thus, the real scale of Qwen's distribution is likely even higher.
Special attention deserves the local deployment segment (GGUF conversions). Here, Qwen also confidently leads: 39.6 million downloads per month versus 20.8 million for Gemma and 7.5 million for Llama. Notably, of the 28,531 Qwen GGUF builds on the platform, only 54 were created by Alibaba itself — the rest are generated by an active community, confirming the deep integration of these models into the daily workflows of developers worldwide.
The Secret to Success: Scaling Strategy and Openness
Qwen's position is explained by three key factors: regular updates to the model lineup, unprecedented coverage of all scales — from compact versions with sub-billion parameters to the giant Qwen3.8-Max (2.4 trillion parameters) — and the Apache 2.0 license, which removes all restrictions on modification and commercial use.
The breadth of the model lineup proved to be a decisive factor. According to Hugging Face data, models with fewer than 1 billion parameters account for 83% of all downloads in the platform's history, while the share of giant neural networks (over 100 billion parameters) is only 1%. Laboratories focused exclusively on large LLMs are losing out: Moonshot AI, which releases almost no models smaller than 70 billion, gathered only 37 million downloads over the year — about 55 times less than Qwen.
Geopolitics of Open AI
The report also records a shift in the balance of power. Almost every month of the current year, the largest open model from China surpassed all US releases in size. The ceiling for Chinese models ranged from 754 billion to 2.78 trillion parameters, while for US competitors, in five out of seven months, it did not exceed 130 billion.
The key difference lies in licensing policy. In China, 59% of models with more than 20 billion parameters are released under Apache 2.0, and 22% under MIT, with none of them having restrictions on commercial use. For US developers, the picture is the opposite: only 29% are under open licenses, 41% under proprietary terms, and 30% without any license specified. Notably, in the US, the leaders in the number of new open models were not AI laboratories but chip manufacturers — AMD and Nvidia, each releasing more than 200 repositories.
My comment: Alibaba's strategy is a masterclass in capturing the open AI market. By betting on accessibility and flexibility, the company is not just increasing downloads but shaping the de facto standard for developers worldwide. American laboratories, which often hide their best developments behind proprietary licenses, will have to reconsider their strategy, or they risk finally losing the initiative in this race.