Apple to power Siri with Google’s 1.2 trillion-parameter Gemini model

Apple Inc. is preparing a significant upgrade to its voice assistant Siri by partnering with Google LLC to use a customised version of its advanced AI model, Gemini, which claims to have roughly 1.2 trillion parameters.

According to reports, Apple would pay Google around $1 billion per year for this licence.

The Gemini model will reportedly handle key Siri functions such as summarising information and planning multi-step tasks, while Apple’s own AI models continue to serve other Siri features.

To protect user privacy and maintain control, the Gemini-based model will run on Apple’s private cloud infrastructure, rather than on Google’s servers. 

This deal is intended as a temporary solution until Apple’s in-house AI systems are powerful enough to take over completely. OpenAI’s ChatGPT and Anthropic’s Claude were also tested before Google's Gemini was finalised for the deal.

Historically, Siri has lagged behind rivals such as Google Assistant and Amazon Alexa in handling complex or multi-stage commands and integrating deeply with third-party services.

With this partnership, Apple aims to narrow that gap and deliver a more capable in-house assistant in spring 2026.

The iPhone-maker is turning to Google’s cutting-edge AI as a bridge to modernise Siri, keeping in-house development ongoing, but recognising it needs external help for now.