Apple Will Power New Siri With Google’s Gemini AI Model
Apple Inc. is reportedly finalizing a deal to use Alphabet Inc.’s Gemini AI model to completely overhaul the Siri voice assistant. Apple is planning to pay Google around $1 billion annually for access to the ultra-powerful technology.
Following extensive testing, Apple chose Gemini over competitors like OpenAI’s ChatGPT and Anthropic’s Claude to serve as an interim solution until its own in-house models are ready. The custom Gemini system has a reported 1.2 trillion parameters, which vastly exceeds Apple’s current models and will significantly boost Siri’s ability to process complex data and understand context.
The new voice assistant, internally code-named Linwood and planned for iOS 26.4, is on track for release next spring.
- Functionality: Gemini will handle Siri’s crucial summarizer and planner functions, enabling the assistant to synthesize information and execute complex tasks. Some basic Siri features will still use Apple’s existing in-house models.
- Data Security: The model will operate on Apple’s Private Cloud Compute servers, ensuring user data remains isolated from Google’s infrastructure.
- Branding: Apple plans to treat Google as a discreet, behind-the-scenes technology supplier, unlikely to promote the partnership publicly. This arrangement is separate from earlier talks about integrating Gemini as a direct chatbot.
Catching Up in AI
The move is an acknowledgment that Apple has fallen behind in the AI race and is relying on outside technology to rapidly advance. The custom Gemini system represents a major leap from the current cloud-based Apple Intelligence model, which has about 150 billion parameters.
However, Apple doesn’t view Gemini as a long-term solution. Despite losing some AI talent, the company intends to develop its own 1 trillion parameter cloud-based model that it hopes to have ready for consumer applications as early as next year, aiming to eventually replace Gemini.
*Cover image credit: Phavitch Theeraphong
[H/T] Bloomberg
