Alibaba on Monday unveiled Qwen 3.5, a new artificial intelligence model it says delivers significantly improved performance at lower cost, as competition in China’s fast-growing AI market intensifies.
The company said Qwen 3.5 is 60% cheaper to use and eight times more capable of handling large workloads than its predecessor. It also introduces what Alibaba calls “visual agentic capabilities,” allowing the model to independently perform tasks across mobile and desktop applications.
“Built for the agentic AI era, Qwen3.5 is designed to help developers and enterprises move faster and do more with the same compute,” the company said in a statement.
The release comes as Alibaba seeks to expand adoption of its Qwen chatbot app in a market dominated by ByteDance’s Doubao and rising AI firm DeepSeek. ByteDance unveiled Doubao 2.0 over the weekend, positioning it as an upgraded model for the AI agent era. Doubao is estimated to have nearly 200 million users in China.
Alibaba said benchmark tests show Qwen 3.5 outperforming leading U.S. models including OpenAI’s GPT-5.2, Anthropic’s Claude Opus 4.5, and Google’s Gemini 3 Pro. The company did not reference DeepSeek in its announcement.
The launch builds on recent efforts by Alibaba to grow Qwen’s user base. Earlier this month, a promotional campaign encouraging in-app purchases through Qwen led to a seven-fold increase in active users, despite technical glitches.
DeepSeek is widely expected to unveil its next-generation model in the coming days, fueling anticipation among investors following the market volatility triggered by its rapid rise last year.
As China’s AI leaders push toward more autonomous “agentic” systems, competition is shifting from chatbot responses to models capable of independently executing complex tasks — raising the stakes in the global AI race.
