Chinese technology giant Alibaba on Monday (16 November) unveiled its new flagship large AI model, Qwen4-Max. The company said the model was better at handling complex, specialised tasks and had significantly stronger safeguards against offensive use.
According to reports from the AliResearch WeChat account, Reuters and other sources, Alibaba’s new foundation model, Qwen4-Max, has 2.4 trillion parameters. The company said Qwen4-Max delivers better performance while using fewer tokens, with continued improvements in key capabilities such as AI agents and coding.
The new model is said to be capable of a wide range of creative and professional work, including 3D modelling, game development, precision manufacturing, finance, law, healthcare and cutting-edge scientific research.
Chinese technology companies hold a prominent position in the global market for open-weight artificial intelligence (AI) models. They are competing fiercely to build more powerful systems while keeping running costs in check.
Alibaba said it screens cybersecurity data according to the specific purpose of each task, retaining data for defensive tasks such as vulnerability verification, detection and remediation, while excluding data for offensive tasks such as enhancing attack capabilities.
The company said the model performs particularly well in avoiding key risks involving injection attacks, memory safety and unauthorised access. It can significantly reduce vulnerabilities in the code it generates by default, helping developers reduce systems’ exposure to attacks from the outset.
A review by Alibaba’s in-house security experts found that Qwen4-Max had substantially improved cybersecurity defence capabilities, as well as significantly stronger safeguards against offensive use. Alibaba also said it had not found evidence that the model could pose large-scale cybersecurity risks.
In recent months, Chinese AI model developers have increasingly disclosed details of safety testing when releasing new models. This comes amid growing global concerns about the risk of AI “going rogue”. AI agents developed by US technology companies have repeatedly been reported to have acted out of control, while an AI agent linked to Chinese technology giant Tencent was reported to have exhibited similar problems in October.