Apple will overhaul Siri in iOS 26.4 using Google’s Gemini on Private Cloud Compute, enabling smarter requests and summaries while keeping data private.
© E. Vartanyan
According to Mark Gurman, Apple has settled on a strategy for revamping Siri, expected as soon as next spring alongside iOS 26.4. Most of the assistant’s new capabilities will run on Google’s Gemini model.
Gemini will execute on Apple’s Private Cloud Compute servers, helping Siri handle user requests. Apple says the updated assistant will be able to answer personal questions by analyzing on-device data and generating responses in real time.
The new Siri architecture will comprise three parts: a request planner, a knowledge retrieval system, and a summarizer. Google Gemini will power the planner and the summarizer on Apple’s servers.
Apple maintains that user privacy will be preserved: Google’s AI will operate solely within Apple’s infrastructure, without sending data to third-party services, while personal data on devices will be processed using Apple’s own Foundation Models.
The new search layer may also draw on Gemini. It is designed to let Siri grasp global topics and respond to general queries without relying on external integrations or a standard web search.
Even though Gemini will be used extensively on the server side, Apple does not plan to promote the partnership. The features will be presented as Apple technologies, running on its servers with a signature interface. The arrangement helps cover gaps in Apple’s in-house LLMs, where current models are not yet mature. The strategy reads as pragmatic—leaning on Gemini where it counts, yet keeping the experience framed as Apple-native. In essence, it resembles how Samsung offers Galaxy AI features on its devices, some of which rely on Google Gemini.