Alibaba jumps 5.4% after unveiling flagship AI model

Alibaba claims its latest AI model is the world’s strongest after Anthropic’s top system. This move comes as China now stands face to face with U.S AI rivals.

The company unveiled a preview of Qwen3.8-Max, the flagship model in its Qwen lineup, through an official X post on Sunday. The preview is now accessible on Alibaba’s Token Plan subscription service and through two of its agent-based developer tools, Qoder and QoderWork.

What makes AQwen 3.8-Max powerful is its 2.4 trillion parameters. Parameters show how much complexity an AI system can handle. The larger the number, the better the performance.

The announcement that came out at Shanghai’s biggest AI event, the World AI Conference, shows a major shift of Chinese developers towards a multi-trillion-parameter target.

BABA jumps for the second time in a week

The news pushed Alibaba’s stock for the second time this week. The Hong Kong-listed shares climbed as much as 5.4% on Monday, putting the company at the top of the Hang Seng Index for the day. Just a couple of days ago, its U.S.-listed shares also jumped about 5% in premarket trading.

That surge happened after Chinese authorities approved Apple’s on-device AI for the Chinese market, as reported by Cryptopolitan previously. The approval had been on hold for some time and links Apple’s iPhone to Alibaba’s Qwen AI for users across the world’s largest smartphone market.

Another Beijing-based startup, Moonshot AI, also made headlines this week. On Thursday, the company launched Kimi K3, a model it says is now the biggest open-source AI system in the world, with 2.8 trillion parameters. This makes it outperform other Chinese competitors like

DeepSeek’s V4 Pro which stands at 1.6 trillion parameters and Zhipu AI’s GLM 5 line at 744 billion. Moonshot said Kimi K3 was purpose-built for tasks like extended coding projects, in-depth research, and complex reasoning.

The company was upfront that K3 still falls short of the strongest closed-access models but said it beat out a number of well-known systems in benchmark tests, among them OpenAI’s GPT-5.5, Anthropic’s Claude Opus 4.8, and Zhipu AI’s GLM-5.2.

Alibaba faces US blacklist and AI theft accusations

The developments for Alibaba come alongside some notable friction. About a month ago, the Pentagon added Alibaba to its list of companies it considers tied to China’s military, placing it alongside automaker BYD and search company Baidu.

Companies on the list are blocked from landing U.S. defense contracts. Defense officials said the move reflects concern about Beijing’s push to bring private sector strength into its military operations.

All three companies rejected the designation. Alibaba said the inclusion had no grounds. “Alibaba is not a Chinese military company nor part of any military-civil fusion strategy,” the firm said in a statement.

On top of that, Anthropic has accused Alibaba of improperly drawing from its Claude AI model, something the U.S. company called a “distillation” attack, where a less advanced model learns by training on outputs from a stronger one. Anthropic said it was the biggest incident of this type it had ever faced.