Focus on performance and lower costs
Alibaba on Monday introduced Qwen3.5, a new artificial intelligence model the company says can independently handle complex tasks while significantly reducing operational costs. The Chinese technology group claims the latest version delivers major efficiency gains, describing it as 60% cheaper to use and eight times more capable at processing heavy workloads compared with its immediate predecessor.
According to Alibaba, Qwen3.5 includes what it calls “visual agentic capabilities,” enabling the system to take actions across mobile and desktop applications without constant user prompts. The company framed the release as a step toward the so-called agentic AI era, in which models move beyond generating text to executing tasks autonomously.
“Built for the agentic AI era, Qwen3.5 is designed to help developers and enterprises move faster and do more with the same compute,” Alibaba said in a statement, emphasizing improved capability relative to inference cost.
Intensifying competition in China
The launch underscores intensifying competition in China’s fast-moving AI landscape. Alibaba is seeking to expand the user base of its Qwen chatbot app, where rivals such as ByteDance’s Doubao and DeepSeek currently hold strong positions.
ByteDance rolled out Doubao 2.0 over the weekend, positioning it as another model tailored to the emerging agentic AI market. Doubao reportedly commands close to 200 million users, making it the largest chatbot platform in China.
DeepSeek, which gained global attention last year with the international release of its models, is expected to unveil its next generation system in the coming days. Anticipation remains high after the company’s earlier breakthrough triggered volatility across global technology stocks.
Strategic push to grow Qwen
Alibaba has recently stepped up promotional efforts to drive traffic to its chatbot ecosystem. Earlier this month, a coupon campaign that encouraged consumers to order food and beverages directly through the Qwen interface reportedly led to a sevenfold surge in active users, despite reports of technical hiccups.
The group has also moved quickly in response to competitive threats. Following DeepSeek’s rapid rise last year, Alibaba released Qwen 2.5-Max, which it said outperformed leading domestic models. In promoting Qwen3.5, Alibaba highlighted benchmark comparisons showing improvements over its previous versions and major U.S. systems including GPT-5.2, Claude Opus 4.5 and Gemini 3 Pro, though it did not reference DeepSeek in the announcement.
Global AI race accelerates
The unveiling of Qwen3.5 reflects broader ambitions among Chinese tech giants to narrow the gap with U.S. leaders in artificial intelligence while differentiating themselves through cost efficiency and integrated consumer ecosystems.
As AI models evolve from conversational assistants into systems capable of acting across applications, the competition is shifting from raw model size to real-world deployment, cost control and user adoption. With fresh releases from multiple players, China’s AI sector appears set for another period of rapid experimentation and rivalry.
