Alibaba Cloud, the cloud computing and AI division of Alibaba Group, is dominating the global open-source AI landscape. Following the release of its Qwen 3.5 model series in February, the company’s models now account for over 50% of all open-source model downloads globally as of March, with total Qwen downloads approaching the 1 billion mark.
Quick Facts
- Qwen nears 1 billion cumulative downloads by March.
- Captured over 50% of global open-source model downloads.
- February downloads were double the next eight rivals combined.
The Qwen Phenomenon
According to a new report by Interconnects AI, a US-based newsletter, Qwen’s download numbers have significantly outpaced competitors like Meta’s Llama and DeepSeek. Citing data from the developer platform Hugging Face, the report highlights that Qwen alone registered 153.6 million downloads in February. This figure is more than double the combined total of the next eight major players, which includes Meta, DeepSeek, and OpenAI.
The surge began last summer when Chinese models overtook their US counterparts on Hugging Face. The release of Qwen models has been a primary driver of this shift, cementing a lead that has left other Chinese players like MiniMax and Zhipu AI with only a fraction of Qwen’s adoption level.
A Small Model Strategy Wins Big
Qwen’s success isn’t just about its flagship models, which Alibaba claims are on par with leading offerings from OpenAI and Anthropic. Florian Brand, one of the report’s authors, attributes Qwen’s dominance to the “extreme popularity” of its smaller model versions.
These models, with fewer than 10 billion parameters, allow developers worldwide to customize and deploy AI solutions at a much lower cost. This accessibility has been a key factor in its rapid, widespread adoption and contrasts with the computationally intensive models from many US tech giants.
Relevance for MENA’s AI Ambitions
This development is particularly significant for the MENA tech ecosystem. With governments and startups across the UAE, Saudi Arabia, and the wider region pouring resources into AI, access to powerful, cost-effective, and customizable models is critical.
Qwen provides a high-performance alternative to dominant US models, empowering MENA-based developers to build and scale AI applications without relying solely on expensive, proprietary systems. As Alibaba Cloud expands its data center presence in the region, the availability of its leading open-source models could accelerate local AI innovation and lower the barrier to entry for new startups.
The Open vs. Closed Source Battleground
The global AI race has a clear fault line between open-source and closed-source strategies. While Beijing has championed the open-source approach to drive broad economic adoption, the US has seen a mix of both. Despite Qwen’s lead, US models like Nvidia’s Nemotron and OpenAI’s GPT-OSS have shown “stickiness” in usage.
However, the trend may be shifting. The immense computational cost of training next-generation models is pushing some players towards proprietary releases. Meta recently launched its Muse Spark model as a closed system after previously championing the open-source Llama series. Similarly, both Alibaba Cloud and Zhipu have kept some of their latest models closed-source to capture revenue directly.
About Alibaba Cloud
Alibaba Cloud is the digital technology and intelligence backbone of Alibaba Group. It offers a complete suite of cloud services to customers worldwide, including elastic computing, database, storage, network virtualization services, large-scale computing, security, management and application services, big data analytics, a machine learning platform, and IoT services.
Source: Tech in Asia
