Apple Intelligence China: Starting November 4, 2026, Apple is testing its first Apple Intelligence features in mainland China. According to the Apple Newsroom (China) post from October 31, the features ship in the latest iOS developer beta, with a public rollout planned for December alongside iOS 27.3. Alibaba separately confirms that its Qwen model (千问) powers selected experiences.
The launch is late relative to other markets: Apple Intelligence shipped in the United States and elsewhere in autumn 2024. China required regulatory filings and a local partner, because western chatbot integrations such as ChatGPT are not available there. Here is what is live — and what is deliberately missing.
Apple Intelligence China: What the beta covers
Apple lists a first wave of on-device and system features embedded in everyday iPhone apps. According to the newsroom, they include:
- Live Translation in Messages, FaceTime, and Phone — including on certain AirPods models
- Writing Tools for rewrite, proofread, and summarize
- Visual Intelligence for identifying objects and places via the camera
- Clean Up in Photos, Genmoji, natural-language photo search, and memory movies
- Priority Mail, email summaries, and smart reply suggestions
- Notification summaries, Reduce Interruptions Focus, and audio/call transcript summaries
- Type to Siri and more natural Siri responses
Importantly, Apple says the update does not include Siri AI or the other next-generation Apple Intelligence features announced at WWDC 2026. MacRumors and TechNode echo that limit; anyone expecting the newer Siri stack will still have to wait.
Qwen inside Siri, Writing Tools, and Visual Intelligence
The main difference from Apple Intelligence elsewhere is the Alibaba partnership. Apple says Chinese users can invoke Qwen directly inside parts of Siri, Writing Tools, and Visual Intelligence for text and image understanding and content generation. Sign-in uses a Qwen account or Sign in with Apple. In a statement carried by Xinhua, Alibaba says it will not use user requests or generated responses for model training and keeps the integration optional.
The pattern mirrors ChatGPT extensions in other regions: on-device features stay Apple’s, while heavier knowledge and generation tasks go to an external partner — here Qwen instead of OpenAI. Apple’s release does not say whether additional Chinese models will be added later.
Devices, languages, and timeline
Apple’s footnotes say the first wave supports iPhone 16 and later, iPhone Duo, iPhone 15 Pro / Pro Max, and iPhone Air, with both Siri and device language set to English and Simplified Chinese. Support for other Apple platforms follows “after the necessary approvals.” Public availability is tied to iOS 27.3 in December; the developer beta begins November 4.
AFP reported on November 4 that the developer trial had begun, with public access still planned for December. On the regulatory side, Xinhua notes that China’s cyberspace authority completed filings in July for several on-device generative services, including Apple Intelligence — a prerequisite for launch.
A sober reading
The rollout closes a gap in Apple’s most important smartphone market, but it does not automatically put the iPhone ahead of local makers that already ship AI assistants. Counterpoint analyst Ivan Lam told AFP the Chinese launch “fills a long-standing gap” without necessarily creating a significant advantage. For developers, the practical point is that the feature set and model partner diverge from the rest of the world; China builds need separate testing.
On code404zero we have covered related model and agent stories, including Meituan’s LongCat-2.5, Ant Group’s Ling-3.1-flash, and frontier-agent measurement in SemiAnalysis GDPx. GitHub Copilot Fusion likewise shows vendors hiding model choice behind routing — similar to how Apple separates on-device work from a cloud partner here.
Regulation and local processing
China’s generative-AI rules generally require filings with the cyberspace authority and domestic data processing. Apple’s newsroom post does not spell out the Qwen plumbing, but it stresses optional enablement and login choice. Alibaba says it will not train on user requests or responses. Whether queries stay in China data centers, and what telemetry Apple itself collects, is only partly public — privacy teams should read Apple’s and Alibaba’s notices side by side.
For China-focused app developers the test matrix changes: features that call ChatGPT elsewhere must be validated against Qwen and the narrower China feature list. Deep links, Writing Tools hooks, and Visual Intelligence flows may diverge even when the UI looks familiar.
Competition and expectations
Local vendors have shipped on-device and cloud assistants for longer. Apple’s launch mainly closes a product gap rather than inventing a category. Omitting Siri AI and the WWDC 2026 stack keeps the release smaller and more approval-friendly — and also lowers the wow factor versus rivals that market cloud agents more aggressively.
Near term, observers should watch beta reliability more than feature count: Live Translation stability, Qwen answer quality inside Siri and Writing Tools, and enforcement of China account rules. Only the December iOS 27.3 release will show whether the developer trial scales to a broad user launch.
What remains open
Apple does not publish detailed beta eligibility beyond the device and language list, nor latency or quality metrics for the Qwen link. Enforcement of account and location rules, and timelines for iPadOS, macOS, or watchOS, are still unclear. There is also no public map of which requests stay on-device versus which go to Qwen.
For product and editorial teams the takeaway is straightforward: Apple Intelligence China enters developer beta today, leans on Qwen rather than western chatbots, omits the newer Siri AI, and aims at the December iOS 27.3 release. Plan against the Apple Newsroom feature list and Alibaba’s opt-in statements — not rumors of “full Siri AI in China.”
