On October 31, 2026, Alibaba and Apple announced a partnership that brings Alibaba's Qwen large language model directly into Apple Intelligence for the Chinese market. The feature opens to testing through the latest iOS developer beta on November 4 and reaches everyday Chinese users with the iOS 27.3 release in December. Through Siri, Writing Tools, and Visual Intelligence, iPhone owners in China will be able to call on Qwen for text and image understanding and content generation. This is the first time a top global consumer-hardware maker has used a Chinese large model as the primary system-level AI backbone inside its own operating system in China — a quiet but striking reversal of the usual "imported AI" default.
Why it is a surprise
For years, the story of on-device AI in China was "Apple builds it, or Apple waits." Instead, Apple is handing the Chinese-language brain of its flagship assistant to a local rival's model. The move reads as a pragmatic surrender: rather than ship a delayed or watered-down first-party model, Apple buys time and regulatory cover by plugging Qwen into the exact surfaces where users notice AI most. Alibaba, in turn, gets a distribution channel measured in hundreds of millions of iPhones — the kind of reach no app-store placement can match.
What Qwen will actually do
The integration is broader than a single button. Siri can draw on Qwen's knowledge and answer questions directly inside Siri, including questions about documents and photos the user chooses to share. The system-level Writing Tools gain a "Compose with Qwen" feature that can draft content on any topic and generate images. On the iPhone camera side, Visual Intelligence lets users point the device at the world and pick "Ask Qwen" to identify what they see. Users can opt in or out, sign in with their Apple ID or their Qwen app account, and decide what to share. Alibaba has pledged that user requests and generated responses will not be used for model training.
The privacy and regulatory angle
Two details make the deal more than a press release. First, the privacy promise: Alibaba says it will not harvest queries or outputs for training, a direct answer to the unease Chinese users feel about sending personal context to cloud models. Second, the regulatory groundwork was laid in July, when China's cyberspace authority filed seven on-device generative AI services for smartphones, Apple Intelligence among them. Without that filing, a system-level AI feature could not have launched at all. The partnership is therefore as much a compliance milestone as a product one.
Why it matters
The signal is that "global hardware plus local model" is now a viable template. If Apple can do it in China, similar splits — brand owned by the device maker, capability licensed from a local lab — may appear in other regulated markets. For Alibaba, Qwen stops being a chatbot and becomes infrastructure inside someone else's ecosystem. For users, the real test is whether Siri-with-Qwen feels like a genuine upgrade or merely a compliant one.
Key data at a glance:
Announced: October 31, 2026
iOS developer beta opens: November 4, 2026
Public rollout: December 2026 with iOS 27.3
Surfaces: Siri, Writing Tools (Compose with Qwen), Visual Intelligence
First: a Chinese LLM as the primary system-level AI backbone in China for a top global hardware maker
Privacy: Alibaba will not use user requests or generated responses for model training
Regulatory: China filed seven on-device generative AI services for smartphones in July, including Apple Intelligence
Sources: Xinhua / China Economic Net, National Business Daily, 36Kr, threerouter
Top comments (0)