Apple Intelligence trial opens in China on Alibaba's Qwen after a two-year delay

This is the first time a Chinese foundation model has been deployed at the operating-system level on a major Western platform. Apple needed a domestic partner to meet Chinese regulations, and chose Qwen after a delay of about two years. Testing started in the iOS 27.3 developer beta on November 4, and a broader rollout is set for December.

Under the arrangement, Qwen powers Apple Intelligence features for users in mainland China. Alibaba has committed that user prompts and model outputs will not be stored or used for training. That commitment is meant to satisfy both Chinese privacy law and Apple's own privacy positioning. 9to5Mac notes that Siri AI, the new third-party-app assistant, is not yet part of the China launch.

Competitively, the deal reinforces a pattern in which Apple sources models from partners rather than building everything itself, much as Samsung relies on Gemini for Galaxy AI. For Alibaba, it puts Qwen on hundreds of millions of devices, at a time when Qwen 3.8 27B is also a community favourite for local inference.

Caveats: the feature set is still limited, Siri AI is missing, and the arrangement depends on regulatory goodwill between the US and China, which has been unstable.