Siri’s overhaul leverages Gemini, under a project that internally renames the model AFM v10 to mask Google involvement.

Google’s Gemini 2.5 Pro currently leads large-language model benchmarks, with ongoing advancement shaping the competitive landscape.

Apple keeps Siri competitive by deploying Gemini as a stopgap while it pushes its own on-device, privacy-first AI under the Apple Intelligence initiative.

The plan includes a near-term Siri upgrade built on Google’s Gemini while Apple works toward internally training a 1-trillion-parameter model expected to be ready next year.

Apple prioritizes user experience over pride, choosing to rely on Google’s technology for Siri’s current generation similar to past decisions to use Qualcomm modems.

Earlier attempts to embed Gemini directly into Siri as a chatbot faced breakdowns from 2024 into early 2025, with future integration still uncertain.

The collaboration extends the existing Google-Apple relationship, applying Google’s AI capabilities within Siri rather than for web search.

Leadership churn in Apple’s AI team, with researchers leaving for competitors, complicates closing the gap with Google.

Apple reportedly struck a temporary deal with Google to power Siri using Gemini due to delays in Apple’s own models.

The partnership aims to deliver Apple Intelligence features—natural dialogue, multi-app automation, and on-device personalization—while Apple builds its own neural models.

China will not use Gemini for Siri; it will rely on Apple’s models and Alibaba’s content filter, with potential partnerships for Baidu’s AI offerings instead.

Apple intends to keep Google as a behind-the-scenes supplier rather than a public marketing partner.