Rival Gpt-5; Alibaba released the QWen3-Max model with a trillion parameter

Chinese company Alibaba has launched its largest and most powerful model of artificial intelligence: QWen3-Max. The model has more than a trillion parameters, and in some benchmarks, it has scored more points from the new Claude and Deepseek models and has even passed the GPT-5 model in the initial rankings. This model was previously introduced in preview mode.

The QWen3-Max Alibaba model is built on 3 trillion token to the optimized architecture of “MOE” that allows it to work optimally despite its large size. This model has two important features that distinguish it from other models.

First: Alibaba has a special focus on the capabilities of the Autonomous Agent in this model. That is, QWen3-Max requires less human commands to achieve a goal set by the user and can make decisions independently.

Second: This model has a stunning power in producing and understanding the code. As shown in specialized benchmarks, this model can solve programming issues with very high accuracy.

Alibaba’s QWen3-Max AI model performance

Alibaba has great claims about QWen3-Max. According to third-party benchmarks, the QWen3-Max has shown a function at the world’s top models. The preview version of the model is ranked third in the LMARNA website and is even higher than the Gpt-5-Chat.

In addition, in the Tau2-Bench benchmark, which measures operating capabilities, the model has gained powerful competitors like Claude Opus 4 and Deepseek v3.1 with a score of 4.3.

Of course, Alibaba will not be satisfied with this model; The company says it is training a heavier and more specialized version called QWen3-Max-Thinking, which focuses on reasoning. This model has also shown extraordinary results and has achieved a full score of 5 % in difficult benchmarks such as Aime 25 and HMMT.

The company had earlier announced that it would invest more than $ 5 billion in artificial intelligence infrastructure over the next three years. Eddie Wu, CEO of Alibaba, emphasized at the company’s recent conference that the company plans to exceed its investment because of the speed of artificial intelligence development.