Alibaba’s Qwen3 surpasses Llama as top open-source model
Alibaba’s Qwen3 model family has surpassed Meta’s Llama as the most-downloaded open-source AI model worldwide, with over 600 million downloads, according to Interconnect‘s 2025 Open Models Year in Review published this week.
The review details that Qwen3 serves as the top choice for fine-tuning tasks, particularly those involving multilingual capabilities. It states that the model family “is regarded as the choice for a lot of problems, especially in terms of multilinguality.” The analysis further confirms that “by now, Qwen has overtaken Llama in terms of total downloads and as the most-used base model to fine-tune.” These metrics position Qwen at the forefront of open-source AI adoption among developers and enterprises seeking customizable solutions.
Alibaba’s consumer-facing Qwen application launched its public beta version on November 17. The app achieved more than 10 million downloads during its first week of availability. By early December, monthly active users across all platforms reached 30 million. This expansion reflects a 149 percent growth rate, establishing the Qwen app as the fastest-growing AI application on a global scale.
The rapid ascent of Qwen has prompted industry observers to describe the reaction in Silicon Valley as “Qwen Panic.” Airbnb CEO Brian Chesky disclosed that his company relies heavily on Qwen models. He characterized them as “very good,” “fast,” and “cheap.” Chesky explained that Airbnb employs 13 different AI models, including those from OpenAI, but noted that ChatGPT’s integration tools were not “quite ready” for production deployment.
OpenRouter’s examination of 100 trillion tokens reveals that Chinese open-source models now constitute nearly 30 percent of global AI usage. This figure marks an increase from 1.2 percent recorded in late 2024. On developer platforms, more than 40 percent of new AI language models created are based on Qwen’s architecture. In comparison, Meta’s share has decreased to 15 percent.
On December 9, Alibaba elevated Zhou Jingren to its elite 17-member Partnership. Zhou serves as Chief Technology Officer for Cloud and leads Tongyi AI Labs, the team responsible for developing Qwen. He spent more than a decade at Microsoft prior to joining Alibaba in 2015. Alibaba recognizes Zhou as the driving force behind Qwen’s development.
Alibaba plans to invest approximately $54 billion over three years in AI infrastructure. AI-related products have recorded triple-digit revenue growth for nine consecutive quarters. The company integrates Qwen throughout its e-commerce, mapping, and local services ecosystems. This deployment positions the AI assistant as a comprehensive platform extending beyond basic chatbot operations.