Apple’s long-delayed new Siri will reportedly “lean on Google’s Gemini” model to power many of its features. It will also include an AI-powered web search.
However, this Gemini integration will supposedly function differently from how Apple currently integrates ChatGPT in iOS.
Gemini made the most business sense for Apple
If everything goes as planned, Apple’s next big Siri update should land in spring 2026. The company’s struggles with developing its own large language models (LLMs) have delayed the release of the next-gen voice assistant.
More recently, Apple’s AI division has been hit with some high-profile exits.
Besides developing its own LLMs, Apple also held talks with Anthropic and Google to power the next big Siri upgrade. The company reportedly created two versions of Siri — one built on its in-house models, and another powered by third-party LLMs.
As part of this process, Apple tested models from both Anthropic and Google, even requesting the latter to develop a custom training model that could run on Apple’s servers.
Bloomberg‘s Mark Gurman, in the November 2nd edition of his Power On newsletter, claims Apple will use Google’s Gemini model to power Siri. Despite Anthropic having a better model, the company went with Google as it “made more sense financially.” The already strong search relationship of the two companies further helped.
Siri’s Gemini upgrade won’t work the way ChatGPT does
There will be one major difference between Gemini’s integration and Siri’s current ChatGPT setup. Right now, Apple clearly hands off complex queries to ChatGPT; you can tell when the switch happens.
That won’t be the case with Gemini. Instead of passing queries off, Apple will use Gemini as the underlying model powering Siri itself. Think of it as Apple quietly tapping into Google’s AI smarts to make its assistant more capable. However, this also means that you should not expect next-gen Siri to mirror Gemini’s feature set on Android.