Apple Taps Google in $1B Deal to Supercharge Siri with Gemini AI

Apple is reportedly finalizing a landmark deal, estimated at $1 billion annually, to license Google’s ultrapowerful Gemini AI. This strategic move is aimed at fundamentally rebuilding the long-underwhelming Siri voice assistant. The partnership will grant Apple access to a custom 1.2 trillion parameter AI model, a staggering leap in complexity compared to Apple’s current in-house technology, which is believed to be around the 150-billion parameter mark. This infusion of advanced AI is expected to power a new slate of sophisticated features for Siri, positioning Apple to finally catch up in the generative AI race.
The decision follows an extensive evaluation period where Apple tested leading models, including OpenAI’s ChatGPT and Anthropic’s Claude. After rigorous testing, Apple zeroed in on Google’s Gemini as the best fit to serve as an interim solution. This alliance underscores an admission from the iPhone maker that its own AI development has fallen behind competitors. The immediate goal is to leverage Google’s cutting-edge technology to bridge the gap while Apple’s own large-scale models, including a rumored 1 trillion parameter version, continue to be developed.
Internally, the effort to rescue Siri using a third-party model is known as project “Glenwood,” spearheaded by high-profile executives including Vision Pro creator Mike Rockwell and software chief Craig Federighi. The completely revamped voice assistant, which is reportedly code-named “Linwood,” is being targeted for a release next spring, potentially as part of iOS 26.4. While the launch is still months away, the finalization of this deal signals a clear direction for Siri’s future, relying heavily on external technology to power its core intelligence.
Under the terms of the arrangement, the 1.2 trillion parameter Gemini model will not take over all of Siri’s functions. Instead, it will be specifically tasked with the most demanding components: the “summariser” and “planner” functions. This means Gemini will handle synthesizing complex information and strategically planning the steps needed to execute multi-part user requests. Other Siri features will continue to run on Apple’s proprietary models, creating a hybrid system designed to balance performance with privacy.
Crucially for Apple, this powerful AI will not be run on Google’s servers. The Gemini model will be hosted on Apple’s own Private Cloud Compute servers, a system designed to ensure that user data is processed in a secure environment and remains completely “walled off” from Google’s infrastructure. This structure allows Apple to leverage a rival’s superior AI without compromising its core privacy promises to its users. The partnership will remain a “behind-the-scenes” arrangement, with Google acting as a technology supplier rather than a public-facing partner, similar to their lucrative Safari search deal but far less visible to the consumer.