Chinese technology Alibaba launched its new artificial intelligence model (IA), Qwen4-Max, with enhanced cybersecurity mechanisms, at a time of growing concern for cyberattacks facilitated by this technology.
© Shutterstock
16/11/2026
by
Lusa
A chinese technology Alibaba launched today its new artificial intelligence model (IA), Qwen4-Max, developed with enhanced cybersecurity mechanisms, at a time when international concerns with cyberattacks facilitated by this technology increase.
The company stated that it deliberately filtered the data used in model training to eliminate offensive cyberattack capabilities, preserving features for computer defense, such as detection and correction of vulnerabilities, according to a statement published on its official website.
Qwen4-Max is a closed model with 2.4 billion parameters in its main architecture.
The number is close to the 2.8 billion parameters of Kimi K3, another of the most advanced AI models developed in China.
According to Alibaba, the new system allows to increase productivity through a greater ability to use professional tools and obtain better quality results.
The company highlighted applications in areas such as three-dimensional modeling, video game development, manufacturing industry, finance, law, health and scientific research.
The IA Arena.ai model assessment platform provisionally placed Qwen4-Max in the first place among Chinese models and in the fifth position of the world ranking, behind systems developed by the U.S. Anthropic, Google, OpenAI and Meta.
According to the artificial analysis company Analysis, the cost of Qwen4-Max is $3.27 (EUR 2.97) per task of its intelligence index, less 67% than the $1.02 (EUR 9.09) of the previous model, Qwen3.8-Max.
The reduction results from both falling prices and lower consumption of tokens, the information units used by AI systems to process and generate responses.
In the field of cybersecurity, Alibaba explained that it kept data for defensive tasks such as verification, detection and correction of vulnerabilities, but eliminated information that could strengthen the system’s offensive capabilities.
“ An assessment by our internal experts concluded that Qwen4-Max’s cyberdefense capabilities have improved substantially and that safety against offensive uses has also been significantly strengthened,” the company said.
Alibaba also announced improvements made after model training to reduce the risks associated with chemical, biological, radiological and nuclear applications.
The launch takes place in a context of growing international debate on the risks of artificial intelligence.
In September, Anthropic officials, Dario Amodei, OpenAI, Sam Altman, and XAI, Elon Musk, defended a slowdown in technology development to ensure security conditions, a position contested by the White House, fearful of losing the technological advantage of the United States over China.
Beijing, for its part, seeks to strengthen international leadership in the sector.
During the summer, Chinese President Xi Jinping advocated “precise and effective” governance of artificial intelligence, while China launched an international cooperation organisation in Shanghai with the participation of about 30 countries.
The Chinese AI sector goes through a period of intense competition, with companies such as DeepSeek, Moonshot, Alibaba, Tencent and Zhipu betting on open models, competitive prices and fast commercial application.
Despite US export restrictions on advanced semiconductors, Chinese systems have been approaching, in terms of performance, the most advanced models developed by US companies.

