Apple's suite of AI utilities, dubbed "Apple Intelligence", will be released in China in December, with Alibaba's Qwen models powering some features, the company announced Saturday.
Chinese users will be able to try out the features as part of Apple's Developer Beta program starting on Nov. 4, with the wider public release planned for iOS 27.3 next month. Alibaba confirmed the partnership the same day, noting that it won't use users' requests to train its Qwen models, many of which are open-weight.
Alibaba has somewhat fallen behind Chinese AI competitors recently, with its latest flagship model, Qwen3.8-Max, ranking 28th among all AI models in The Latent's benchmarks with a Latent Score of 96.6. Moonshot AI's Kimi K3.1 Pro is currently the highest-ranked Chinese model at #7 with a score of 136.0, the only Chinese model currently in the top 10.
Siri will route some requests to Qwen
After the update, Siri will be able to pass some questions to Qwen, including when a user asks about the contents of a photo, PDF or slide deck on their screen. Apple's AI-powered Writing Tools will get a Compose option that uses Qwen to draft text from scratch, and Apple's Visual Intelligence, which identifies things seen through the camera's lens, will also tap the model. To use the features, users have to sign in with a Qwen account or with Sign in with Apple.
Users will be able to choose to turn on Qwen and limit what data is shared with it, Alibaba said. The tech and e-commerce firm's CEO, Eddie Wu, said Qwen would be integrated deeply into Apple's products, according to state news agency Xinhua. The features require an iPhone 15 Pro or newer that's set to either English or Simplified Chinese.
The features that will be powered by Qwen in China are generally powered by OpenAI's ChatGPT models in the rest of the world, though users don't need to sign in to a ChatGPT account to use them.
The China launch of Apple Intelligence will also include other features not powered by Qwen, including Apple's live translation feature, AI-generated "Genmoji," a tool to remove unwanted objects in photos, and auto-generated summaries for emails and app notifications, which all run on Apple's own foundation models. The overhauled and AI-powered version of Siri, which was unveiled at Apple's WWDC 2026 conference and began rolling out in September with the release of iOS 27, will not come to China with the December update, however. Apple said in September it's still working through getting through regulatory requirements for it.
Chinese regulators signed off in July
The Cyberspace Administration of China (CAC), the country's internet regulator and censor, listed Apple Intelligence among seven on-device generative AI services that had completed its filing process, per its July 15 notice, alongside offerings from Huawei, Xiaomi, OPPO, vivo, Samsung and ZTE. The December launch will only cover iPhones, with Apple noting that unspecified other platforms will follow once it gets the necessary approvals.
Wang Xiaojie of Beijing University of Posts and Telecommunications said the partnership marks the first time a major global consumer hardware maker has picked a Chinese model as its main system-level AI in the country, per Xinhua. Beijing last month cleared DeepSeek and MiniMax to deploy their models outside China.
Apple, led by new CEO John Ternus as of September, has mostly shied away from building its own frontier models, choosing instead to partner with OpenAI and, reportedly, Google for various AI features. Apple also worked with Meta in October to optimize its Muse Spark Open model so it can run locally on Macs, The Latent previously reported.
