Apple is nearing a deal with Google to pay approximately $1 billion annually for a custom version of Google’s Gemini AI model, which will power an overhaul of the Siri voice assistant, according to a Bloomberg report.
Apple has historically developed its own AI technology in-house. The company intends to integrate Google’s custom model as an interim measure until its proprietary AI achieves sufficient capability to support a series of planned features for the voice assistant. This arrangement marks a departure from Apple’s self-reliant approach to AI development.
The custom Gemini model features 1.2 trillion parameters, a metric that quantifies the software’s complexity and performance potential. In comparison, Apple’s existing cloud-based Apple Intelligence model operates with 150 billion parameters. This makes the Google model approximately eight times more complex than Apple’s current offering, enabling enhanced functionality for the Siri upgrade.
Earlier this year, Apple assessed AI models from OpenAI and Anthropic as potential options. The company conducted tests on models from these providers alongside Google’s. Following this evaluation process, Apple selected Google’s Gemini for the Siri integration.
Bloomberg indicates that the revamped Siri is scheduled for release in the spring of next year. With the launch several months in the future, the development timeline remains subject to potential adjustments based on ongoing progress.
