Alibaba Launches 'Qwen 3.5' Model to Keep Up with the Era of AI Agents

SadaNews - On Monday, Alibaba revealed its new artificial intelligence model "Qwen 3.5," designed to autonomously perform complex tasks with significant improvements in performance and cost, with the Chinese tech giant claiming that it surpasses major American competitors' models in many criteria.

This release comes as "Alibaba" seeks to attract more users to its "Qwen" chatbot application in China, a market currently dominated by the "Doubao" chatbot - owned by ByteDance - and "DeepSeek," whose developer became the first Chinese AI company to achieve a global breakthrough last year.

"Alibaba" stated that "Qwen 3.5" is 60% cheaper to use and eight times better in handling large workloads compared to its predecessor, adding that the model also possesses the ability to take independent actions via mobile and desktop applications, or what the company calls "visual agent capabilities."

The company said in a statement: "Qwen 3.5 is designed for the era of agent-based AI, to help developers and businesses move faster and accomplish more with the same computing power, setting a new efficiency benchmark for each unit of inference cost."

This launch follows ByteDance's announcement on Saturday of "Doubao 2.0," an update to its chatbot application, which currently has the largest user base in China, nearing 200 million users. Similar to Alibaba's announcement, the new model has been promoted as suitable for the era of agent-based AI.

The launch of "Qwen 3.5" could contribute to enhancing the gains that "Alibaba" has recently achieved amid intense competition among AI models in China.

Earlier this month, an initiative by the e-commerce giant, distributing vouchers that encouraged consumers to purchase food and drink directly through the Qwen app, resulted in a sevenfold increase in active users, despite some technical issues.

Last year, "Alibaba" was one of the first competitors to "DeepSeek" to quickly respond to the startup's rise, launching "Qwen 2.5-Max," which it claimed surpassed one of "DeepSeek's" successful models.

The company did not mention "DeepSeek" technology in its announcement of "Qwen 3.5," with the benchmark test results it published only showing the new model's superiority over its previous version and competing American models "GPT-5.2," "Claude Opus 4.5," and "Gemini 3 Pro."

"DeepSeek" is expected to launch its new next-generation model in the coming days, increasing anticipation among investors and industry insiders, especially after the decline of global tech stocks that the company caused a year ago upon the launch of its "R1" model.