Apple has registered its artificial intelligence platform, Apple Intelligence, for use on iPad and Mac in China, a requirement before a public launch.
The move keeps Apple on track for a December debut with iOS 27.3, Softonic reported. This follows July's regulatory approval for the iPhone, finishing a rollout delayed by months of negotiations.
The Qwen Model Deal
Apple cannot use the same infrastructure or partners in China as in other markets. Instead, the company will rely on Alibaba’s Qwen model, several outlets confirmed. Baidu is also cited as a collaborator.
This partnership is a direct result of Chinese regulations that require generative AI services to register with the government and use local infrastructure and companies. Apple must comply with Chinese law while maintaining a consistent user experience across markets.
“Apple has been forced to adapt its strategy to the particularities of China, to the point of renouncing the Google models it uses elsewhere.”
Applesfera noted the move results in a different version of Apple Intelligence, shaped by the country's regulatory demands.
A Phased Rollout in China
The deployment in China will be phased. The initial wave in December with iOS 27.3 will include basic features like text summarization, real-time translation during calls, and visual recognition. However, the more advanced conversational Siri capabilities will be held back.
This staged launch reflects a delicate market for Apple, facing strong competition from Huawei and renewed user concerns about data privacy and control due to the reliance on local partners and on-device processing.
Developers in China have already begun testing. The first beta for iOS 27.3 was released in early November, allowing for preliminary trials before the public release, Softonic reported. The Korea Herald, as noted by NeoTeo, confirmed that developer tests for the Qwen integration began on November 4.
Users Control Data and Grant Permission for Qwen Integration
A key detail for users is the level of control they will have over their data. Asia News Network reported that users can activate or deactivate the Qwen integration and will be prompted for permission before sending a query, image, or document to the Alibaba model.
They can use either their Apple ID or a Qwen account for login. The same publication attributed to Alibaba the statement that requests and responses from this integration will not be used to train its models. In practice, content will only be sent to Qwen after a user grants explicit authorization.
Apple Intelligence Requires Specific Devices and Regional Settings
Not all devices will be eligible at launch. According to iPhoneA2, Apple Intelligence will be available on the iPhone 15 Pro, iPhone 15 Pro Max, the iPhone 16 family, and subsequent models, including the iPhone Air and Duo. Both the device system and Siri must be set to English or Simplified Chinese.
Access also requires the device location and Apple ID account to be configured in China. Support for iPad and Mac will come after the necessary approvals are obtained, though an exact public launch date for those platforms is not yet known.
The contrast with other markets is stark. While Apple Intelligence in most of the world now uses Google’s Gemini models, in China it is built on Alibaba’s Qwen. This dependency on third-party models may explain why John Ternus, Apple’s CEO, has reportedly become more insistent on developing stronger in-house models for the future, Applesfera suggested.
