Alibaba Challenges Meta in AI with New Laptop-Ready Qwen Models

Chinese tech giant Alibaba has escalated its artificial intelligence rivalry by launching Qwen3.8-27B, an open-weight AI model capable of running on laptops. Concurrently, the company also released the weights of its most powerful AI model, Qwen3.8-Max, making them accessible to developers as open-source. This strategic maneuver intensifies competition with rivals like Meta Platforms Inc. (META) in the open-weight AI market and is part of Alibaba's broader effort to solidify its leadership in this rapidly evolving sector.

Alibaba's new Qwen3.8-27B model features 27 billion parameters and offers robust capabilities in coding, professional tasks, research, and long-horizon agentic tasks. The company claims this model performs comparably to models ten times its size. Meanwhile, the Qwen3.8-Max model, with 2.4 trillion total parameters and 95 billion active parameters, stands as Alibaba's most capable AI model to date. Its open-sourcing marks the first time a model of this scale has been made publicly available. This development follows Meta's recent introduction of Muse Glimmer, a 30-billion-parameter open-weight model, and its stated intention to open-source the weights of its flagship Muse Spark 1.2 model.

The on-device AI market is rapidly becoming a crucial battleground for model developers. According to industry analysts, locally run models can deliver quicker responses and keep data off remote servers, offering a significant privacy advantage for some users. Alibaba's Qwen-based models have already achieved considerable success in the open-weight AI domain; Hugging Face reports indicate that derivative works built on Qwen models have reached 151,448, approximately 2.6 times Meta's total. Furthermore, over the past six months, Alibaba's Qwen open models have accumulated more than 3 billion global downloads, surpassing Meta (227 million) and Google (418 million).

These actions should be viewed within the broader context of the intensifying US-China AI race. Chinese firms, including Alibaba, Moonshot AI, and DeepSeek, are gaining significant global traction by offering capable models that are relatively inexpensive and easy to customize. Despite chip export controls and AI restrictions, China's progress in this field appears undeterred. Open-source models provide developers with building blocks for creating new AI products, boosting adoption rates and serving as a key metric of influence in the US-China AI competition.

Analysts and market observers suggest that competition in the open-weight AI space is increasingly centered on which companies can offer the most capable models and superior on-device performance. Nick Patience, AI lead at the Futurum Group, noted that Meta's re-embrace of open weights was a direct response to the significant strides made by Chinese labs over the past two years. The release of smaller, open-weight models like Qwen3.8-27B makes it more feasible for developers to run these models on their own hardware, prompting new discussions around privacy and control. As these models become more widespread, a broader adoption of AI technologies across various sectors is anticipated.