By leveraging Google’s Gemini through distillation, Apple is building powerful, efficient AI that can run directly on its devices—marking a major step toward private, on-device intelligence at scale. (Source: Image by RR)

Siri and System Features Could Benefit From Advanced AI Capabilities

Apple has reportedly secured extensive access to Google’s Gemini models, giving it the ability to use one of the world’s most advanced AI systems inside its own infrastructure. The deal, according to an article in the-decoder.com, allows Apple not only to run Gemini in its data centers but also to apply a powerful technique known as distillation—training smaller, efficient models based on Gemini’s outputs.

Through distillation, Apple can generate high-quality responses (and reasoning steps) from Gemini, then use that data to train lightweight models that mimic its performance. These smaller models require far less compute and can run directly on iPhones, iPads, and Macs—aligning with Apple’s long-standing focus on on-device processing and privacy.

This approach effectively lets Apple leverage frontier AI without relying entirely on cloud-based systems. While Gemini is optimized for chatbot and enterprise use cases, Apple is expected to adapt distilled versions for its own ecosystem, particularly for Siri and other system-level features. At the same time, Apple continues developing its own in-house models through its Foundation Models team.

The strategy positions Apple to compete in AI without building everything from scratch—combining external cutting-edge models with its own hardware-software integration. With new AI features likely to be unveiled at Apple’s upcoming developer conference, the company appears poised to make a significant leap in delivering powerful, private, and efficient AI experiences directly on its devices.

read more at the-decoder.com