A report details the financial agreement between Apple and Google for the integration of Google's Gemini AI model into Apple products.

Apple will reportedly pay Google a large sum to use a custom version of its Gemini LLM model while it works on its own.
According to a report, Apple intends to use a customized 1.2 trillion parameter AI model of Google's Gemini Large Language Model. Apple and Google are reportedly nearing an agreement where Apple would pay Google $1 billion annually for the right to use Gemini's LLM. The report cites unnamed sources familiar with the private discussions between the two tech companies.
The model's size reflects the AI software's intricacy and exceeds the size of Apple's current models. Apple plans to use Google's Gemini until its own models are sufficiently advanced. Apple tested models from other AI providers, like OpenAI’s ChatGPT and Anthropic’s Claude. However, earlier in the year, Apple determined that Google's Gemini was the optimal choice, and the companies are working to finalize a deal.
The 1.2 trillion parameter Gemini AI model that Apple wants to use with Siri is far larger than the 150 billion parameter model used with the cloud-based version of Apple Intelligence, an AI initiative that has drawn some criticism. The larger model should enable Apple Intelligence to be more effective, understand context better, and handle complex data. Siri is intended to be the public face of Apple's AI efforts, and delays have reportedly pushed the launch of Personal Siri to iOS 26.4 this spring, an update internally codenamed Linwood.
The report indicates that within Apple, the plan to use a third-party AI model to enhance Siri is referred to as Glenwood. Mike Rockwell, the creator of the Vision Pro headset, and Craig Federighi, the head of software engineering, are managing the project.
Apple will utilize Gemini to assist Siri in synthesizing information and determining the optimal approach to handle intricate tasks. The report specifies that some of Siri's capabilities will continue to rely on Apple's own models. To safeguard data from Apple device users from Google's infrastructure, the Gemini model will operate on Apple’s Private Cloud Compute servers. The report says that Apple has already allocated AI servers to run the model.
Apple anticipates eventually developing its own model to replace Gemini. The company is reportedly developing a 1 trillion parameter cloud-based model for consumer use as early as next year. This endeavor may prove challenging for Apple, given recent departures of AI personnel. Currently, Google's 2.5 Pro version of Gemini is considered the leading LLM.
With AI becoming a key feature in smartphones, Google has arguably used AI most effectively among smartphone makers to create useful features. Using Gemini in place of Google Assistant offers exceptionally detailed answers. Gemini can also set alarms and timers.
Another major partnership between Apple and Google involves the latter's dominant search engine. Court documents revealed that Google pays Apple 36% of advertising revenue from search requests on Apple devices to be the default search engine on Safari. In 2022, this deal resulted in a $20 billion payment from Google to Apple. The new Gemini partnership may not be advertised in iPhone ads.