As part of the Apple-Google AI partnership, the tech giant will reportedly use a massive Google AI model for Siri and its new “Apple Intelligence” features. This model is claimed to have 1.2 trillion parameters and is “exorbitantly expensive” even for Google.

Details of the Apple-Google AI Deal

Bloomberg’s Mark Gurman reports that Apple plans to use this massive, custom-built Gemini AI model to power Siri. This Gemini model will process more complex user requests by offloading them to the cloud. Apple will also introduce its Private Cloud Compute framework, which uses encrypted and stateless data to protect user privacy. The company also tested large language models like OpenAI’s ChatGPT and Anthropic’s Claude before making its final decision.

According to Gurman, Apple will pay approximately $1 billion annually to use Google’s proprietary AI technology. This is just the latest in an ongoing transactional relationship between the two tech giants. However, Google already pays Apple $20 billion annually to secure default search engine privileges in the Safari browser and other Apple services. Apple has codenamed its efforts to refine Siri with the help of third-party models “Glenwood.” These efforts are led by Vision Pro title creator Mike Rockwell and software engineering chief Craig Federighi.

Under Apple’s evolving architecture, this proprietary Gemini model, acquired from Google, will handle Siri’s “query planner” and “summarizer” functions. The “information search” system will be left to Apple’s own on-device large language models (LLMs).

However, this partnership will not place Google’s search AI at the forefront of the Apple ecosystem. The Gemini model will provide critical support for the revamped Siri, which Apple plans to launch with iOS 26.4, featuring features like in-app actions, personal context, and screen awareness.

So, what are your thoughts on Apple’s use of Google’s AI model for Siri? Share your thoughts in the comments!