The latest news suggests that Apple would be preparing a historic change: the new Siri would rely on a Google Gemini custom model to offer much more advanced AI functions from March 2026. This strategy would fit with Apple's plan to relaunch its assistant after years of minor iterations and with a focus on privacy, integrating the heaviest part of the processing in Private Cloud Computing and maintaining control of the user experience.
The movement would not be isolated. In recent months, Tim Cook has reiterated that the “improved” Siri is still on track for 2026 and various leaks from the specialized press speak of an agreement by which Google would develop a bespoke Gemini for Apple, executed on the Cupertino company's infrastructure. For the user, the interface would still be “Siri”, but with more natural responses, context memory and new navigation capabilities. agent to understand complex requests.
What would change on a daily basis with a Gemini-based Siri?
If Apple finally integrates a custom Gemini model, we can expect tangible improvements on three fronts. First, understanding and conversation: a Siri capable of following the context, resolving ambiguities and asking for clarification when necessary, with a more natural and less robotic tone. Second, compound tasks: chain actions (“organize my trip photos, create an album with the best ones and share it with the family”) without the user having to guide each step. And third, more complex actions within the applications, thanks to an expanded set of intentions: open documents, edit notes, extract data from an email or modify several system settings in a single command.
Apple would also have room to enhance functions of summary, translation, transcription and text generation at the system level, with Siri as the gateway. All this while maintaining its discourse focused on privacy: an “encapsulated” Gemini that runs on servers controlled by Apple, with anonymization and audits, and in parallel to the models on device for quick tasks when there is no connection.
How it fits with Apple Intelligence and iOS calendar
The pieces fit with the progressive deployment of Apple Intelligence. The most repeated roadmap speaks of a launch of the Siri renewed in March 2026foreseeably associated with an intermediate version of iOS with improvements focused on the assistant. That window would allow Apple to mature the orchestration layer and expand compatibility to more languages and regions.
At an ecosystem level, it wouldn't be surprising to see a more useful Siri in the HomePod, the Apple TV and in a device with screen and speaker for the home, where voice control makes real sense. On the iPhone, the key will be for the new AI to be noticed without learning commands: speak naturally and get results.
And the buts? Risks, dependency and expectations
The main question is in technological dependency: To what extent can Apple delegate to a Google model without compromising its control? If the agreement is finalized, the reasonable path is a development tailored and with limits: Gemini as a statistical engine, but under rules, filters and telemetry parameters defined by Apple. Another risk is that of the expectations: After years of promises, the new Siri will have to demonstrate reliability, speed and language coverage. And, of course, we will have to see What functions come to each country and what compatible devices.
A chance for Siri to be relevant again
If everything comes true, 2026 may be the year in which Siri once again competes directly with Google Assistant and other assistants based on generative models. The key will be to execute the plan well: real privacy, clear utilities and a uniform experience throughout the ecosystem. After the presentation of the iPhone 17Siri is expected to become the main topic in the coming months. Do you think Apple should rely on Gemini to accelerate this leap in quality? We read you in the comments.
