Apple trained a China-specific AI model with Alibaba, regulatory filings confirm
Chinese network-access filings completed 29 September registered Apple's own model separately from Alibaba's Qwen and Wanxiang, which Apple also offers as options.
- Government & policy
- Models & capabilities
- Notable
Apple trained its own large language model for the China market, working with Alibaba on the underlying training and on meeting China’s content and compliance requirements, Reuters reported, citing unnamed sources. The report, carried by MacRumors and other outlets, described Apple as the first foreign company approved by the Chinese government to offer a proprietary AI model in the country — rather than being required, as other foreign firms operating in China have been, to license and deploy a domestic model outright.
According to the reporting, Apple pursued a dual-track approach: its own proprietary model sits alongside a separate integration of Alibaba’s Qwen, which Apple was expected to offer as an optional assistant on iPhones, iPads, Macs and Vision Pro devices sold in China, similar to how it offers ChatGPT as an option elsewhere. Reuters reported that China’s Cyberspace Administration had logged Apple’s generative AI service the previous month, a registration step required before any public rollout; at that time, Alibaba itself had already confirmed to reporters that Qwen would be integrated into Apple Intelligence across iOS, iPadOS, macOS and visionOS in China, though without a launch date. Apple had also briefly published — then withdrawn within a day — a support document describing how to connect Siri to Qwen, which some outlets took as accidental confirmation of the arrangement ahead of the report. Apple had reportedly evaluated Baidu and DeepSeek as partners before settling on Alibaba.
None of this had been confirmed by Apple itself at the time of reporting, and no public rollout date had been set; Reuters described a wider release as expected “in the coming months.” The report was nonetheless notable as a sign of how a foreign company with no China-approved model of its own had navigated Beijing’s generative-AI registration regime — building its own system rather than defaulting entirely to a Chinese partner’s, while still relying on that partner for compliance and for a fallback assistant option.
The dual-track account held up. Updated filings with China’s Ministry of Industry and Information Technology, completed 29 September and surfaced publicly on 14 October, registered three separate systems for network access covering the iPhone 15 Pro and later models: Apple Intelligence itself under its own registration number, alongside separate entries for Alibaba’s Tongyi Qianwen (Qwen) language model and its Tongyi Wanxiang image-and-video model. The filings did not by themselves establish a public launch date, but they confirmed that Apple’s own system and Alibaba’s models had each cleared a required Chinese regulatory step as distinct, separately registered services rather than one being folded into the other.
The following day, Apple chief executive Tim Cook made the first company-level confirmation, telling Chinese media in Shanghai that a test version — covering live translation, writing tools and photo cleanup, but not the new Siri AI — was due by the end of October.