Apple is preparing to overhaul Siri by integrating a custom version of Google’s Gemini artificial intelligence model, according to a report from Bloomberg. The iPhone maker is expected to pay Google around $1 billion per year for access to the AI, which features 1.2 trillion parameters, a measure of the model’s complexity and ability to handle tasks.
The adoption marks a notable shift for Apple, which has historically relied on its own AI technology. The tech giant plans to use Google’s model as an interim solution while it develops more advanced in-house AI. Apple’s current cloud-based AI infrastructure operates with 150 billion parameters, making Gemini roughly eight times more complex.
Apple evaluated models from OpenAI and Anthropic earlier this year but ultimately selected Google’s solution. Cost considerations reportedly influenced the decision, with Anthropic’s fees seen as higher than Apple’s target. The company’s existing partnership with Google for search results, which generates approximately $20 billion per year for Apple, may have also reinforced the collaboration.
The custom Gemini model will run on Apple’s Private Cloud Compute servers, ensuring that Google does not have access to Apple’s data. It will handle functions such as multi-step task execution, summarization, and other complex queries across iOS. Apple’s own AI models will continue to support certain Siri features, indicating a hybrid approach during the transition.
Gemini employs a Mixture-of-Experts architecture, activating only a fraction of its parameters per query. This approach allows the model to provide high computational power without excessive processing costs.
Apple originally intended to debut the updated Siri with iOS 18, but performance deficiencies prompted a major rebuild of the assistant’s architecture. The revised Siri, referred to internally as Apple Intelligence, is now expected to launch in spring 2026 with the iOS 26.4 update. The updated assistant is designed to handle more complex questions, execute multi-step tasks, and operate across multiple apps more effectively. While it will resemble chat-based AI platforms like Claude or ChatGPT in functionality, Apple does not plan to release a standalone chatbot.
Apple is continuing work on its own large-scale AI models, including a cloud-based system with around one trillion parameters, which could be ready as early as 2026. Once the in-house technology reaches sufficient capability, Apple plans to transition Siri to its proprietary AI infrastructure, reducing reliance on Google.
The move underscores Apple’s broader AI strategy, balancing immediate product improvements with longer-term ambitions to maintain control over core technologies and user data. Analysts suggest that leveraging Google’s Gemini allows Apple to accelerate Siri’s capabilities without compromising privacy standards while continuing its internal AI development.
Notes: This post was edited/created using GenAI tools. Image: DIW-Aigen.
Read next: YouTube Deletes Palestinian Rights Videos, Complying with U.S. Sanctions that Shield Israel
