Alibaba Group has unveiled a preview version of its next generation flagship artificial intelligence model, Qwen3.8 Max, as the Chinese technology giant seeks to strengthen its position in the rapidly evolving global AI market. The announcement was made during the World Artificial Intelligence Conference (WAIC) in Shanghai, where companies showcased the latest advances in frontier AI models and applications.
The new model, named Qwen3.8-Max-Preview, features approximately 2.4 trillion parameters, making it the largest model in Alibaba's Qwen family to date. According to the company, the model is designed to compete with leading frontier AI systems and will be made available as an open-weight model in a future release, allowing developers and enterprises to download, customise and build applications on top of it.
Alibaba said the preview version is currently available through its AI development platforms, including Token Plan, Qoder and QoderWork, where developers can begin testing its capabilities. The company described the model as one of its most advanced AI systems, capable of handling complex reasoning and enterprise-grade workloads. However, independent benchmark evaluations of Alibaba's performance claims have not yet been published.
The latest launch comes amid an increasingly competitive AI landscape in China. Last week, Beijing-based Moonshot AI introduced its Kimi K3 model with 2.8 trillion parameters, while several other Chinese AI companies, including Z.AI and MiniMax, have also released increasingly capable large language models in recent months. These launches reflect the growing pace of AI development as Chinese firms seek to narrow the gap with global leaders such as OpenAI, Anthropic and Google.
Industry analysts say the race to develop frontier AI models is becoming more resource intensive, requiring significant investments in computing infrastructure, data and specialised talent. As a result, the market may gradually consolidate around a smaller number of companies capable of funding large-scale model development while simultaneously building commercial ecosystems around their AI platforms.
Alibaba has been steadily expanding its AI portfolio over the past year. Earlier in 2026, the company introduced new multimodal, image generation and agentic AI models while reorganising its AI businesses under a dedicated division to accelerate commercialisation of its artificial intelligence offerings. The company has also increased investments in enterprise AI services delivered through Alibaba Cloud, positioning the Qwen family as the foundation of its broader AI strategy.
The decision to eventually release Qwen3.8 as an open-weight model aligns with Alibaba's broader approach of encouraging developer adoption and enterprise experimentation. Open-weight models have gained popularity because they enable organisations to deploy AI systems on their own infrastructure, adapt them for specialised use cases and reduce dependence on proprietary cloud-hosted services.
The preview also underscores China's accelerating push in generative AI as domestic technology firms continue investing in increasingly sophisticated models. While companies continue to compete on benchmark performance and parameter counts, industry observers note that future success will increasingly depend on practical enterprise adoption, developer ecosystems and the ability to deliver reliable AI applications at scale.