Alibaba Unveils ”Qwen 3.8 Max”, Its Largest and Most Powerful AI Model to Date
Alibaba unveiled on Monday what it described as its largest and most powerful artificial intelligence model developed to date, dubbed "Qwen 3.8 Max". The new release closely rivals the scale of a model launched last month by its Chinese competitor, Moonshot AI.
Chinese technology companies, which represent a major global force in developing open-weights AI models, are currently engaged in a fierce and rapidly accelerating race to build more powerful models without incurring prohibitive operational costs.
Qwen 3.8 Max boasts approximately 2.4 trillion parameters—the numerical values the model learns from data to recognize patterns, generate responses, and execute tasks. This scale is comparable to the 2.8 trillion parameters featured in Moonshot AI's Kimi K3 model.
Computing Power and Data Volumes
While a higher parameter count does not automatically guarantee a superior model, it has emerged as a crucial indicator of the massive computing power and data volume utilized in the development of advanced AI systems.
Chinese tech firms are particularly keen on publicly announcing the parameter counts of their models to drive broader adoption within the developer community. The vast majority of their models rely on an open-weights approach, which allows developers to download the core learned values to run or actively modify the models themselves.
Open-Weights vs. Closed-Source
In stark contrast, leading Western tech giants such as OpenAI, Anthropic, and Google typically do not disclose the parameter counts for their closed-source AI models.
Alibaba introduced Qwen 3.8 Max via the AI Arena platform, a benchmarking site specializing in comparing AI models. The model immediately secured the position as the highest-ranked Chinese model in the text generation category. However, it currently still trails behind Claude Fable 5 and three versions of Opus, all of which were developed by Anthropic.