07 Nov 2025
Apple has reportedly finalized a deal to license a custom 1.2 trillion‑parameter version of Google’s Gemini to power a long‑planned overhaul of Siri, committing roughly $1 billion annually, according to the Rundown summary of Bloomberg reporting. The plan would put Gemini behind Siri for tasks like summarization and multi‑step planning, with inference occurring on Apple’s Private Cloud Compute infrastructure to protect user data.
Apple evaluated models from OpenAI and Anthropic during trials; the chosen 1.2T configuration is substantially larger than the ~150B parameters used in Apple’s current Apple Intelligence model. Bloomberg says Apple intends for Google’s role to be “behind‑the‑scenes” rather than a public partnership. The new Siri experience could arrive as soon as next spring, with Apple viewing Gemini as a temporary stopgap while it continues to develop its own internal model.
Why this matters: after years of delays and internal challenges, adopting Gemini would quickly bring more capable LLM capabilities to Siri, but it also raises questions about long‑term independence as Apple attempts to build its own AI solution amid hiring and retention headwinds.
Source