Apple has trained a large language model built specifically for the Chinese market, working closely with Alibaba to get it done, according to a Reuters exclusive published Friday citing three people familiar with the matter. It’s a notable shift for a company that had, until now, planned to lean entirely on Chinese partners’ models to bring AI features to iPhones sold in China.
The people, who spoke on condition of anonymity because the information isn’t public, said the model was developed with Alibaba’s direct support. Reuters noted this is the first time Apple’s work on its own China-specific model has been reported. Apple and Alibaba did not respond to requests for comment.
Why Apple Needed a Plan B
China is the one major market where Apple can’t simply ship its existing AI stack. Anthropic’s Claude and OpenAI’s ChatGPT — the models Apple pairs with its own technology everywhere else — aren’t available in the country. So Apple has spent the past year building a workaround with local partners instead.
Under an arrangement approved by China’s Cyberspace Administration last month, Alibaba’s Qwen model was set to be built into the version of Apple Intelligence shipping on iPhones, iPads, Macs, and Vision Pro devices sold in China. Reuters reported in July that Baidu’s technology was also part of that plan. Apple’s Alibaba partnership was first confirmed back in February.
Training its own model on top of that arrangement gives Apple something it didn’t have before: more direct control over how its AI actually behaves in China, rather than depending entirely on outside vendors. The timing matters — Apple is fighting hard to hold ground against Huawei and other domestic phone makers whose on-device AI features have improved rapidly.
A Guide That Appeared, Then Vanished
One detail in Reuters’ report stands out. Last week, Apple published a Chinese-language support guide explaining how eligible Mac users in mainland China could connect Alibaba’s Qwen to Siri and Apple’s Writing Tools. Apple then deleted the guide, without saying why.
It’s a small thing on its own, but it fits a pattern: Apple is clearly moving pieces into place for a China AI launch, testing and pulling back documentation as it goes, ahead of an official rollout.
When Does This Actually Launch?
Apple Intelligence is expected to arrive in China in the coming months through an iOS update, with earlier reporting suggesting the rollout could line up with new iPhone launches this September. No firm date has been confirmed by Apple.
The Reality Check
This story is built entirely on anonymous sourcing — “three people familiar with the matter” — which is standard for Reuters exclusives of this kind, but it means Apple and Alibaba haven’t confirmed a single detail publicly. Treat specifics like model size, training data, or exact launch timing as unconfirmed until either company goes on record. We’ll update this piece if that changes.
Why This Matters Beyond China
If Apple is willing to build a dedicated, market-specific AI model just to satisfy local regulatory and competitive pressure in China, it raises an obvious question for every other market where Apple operates under different data rules — including India, where the debate over data localisation under the DPDP Act is still playing out. Apple has never built an India-specific model; for now, it ships the same Apple Intelligence stack backed by third-party partners globally. Whether regulatory pressure elsewhere ever pushes Apple toward a similar market-specific build is worth watching, especially as more countries tighten rules around where AI training and inference actually happen.
For now, China remains a special case — the only major market where Apple can’t use its usual AI partners at all. But the playbook it’s building there, model ownership plus a local compute partner, is one other regulators may eventually ask for too.
Source: This report is based on an exclusive by Reuters, citing sources familiar with the matter. Apple and Alibaba have not officially confirmed these details.