Apple partners with Google to use Gemini models for revamped Siri on iPhones

Under the agreement, Gemini will power the revamped Siri that Apple had demonstrated back at WWDC 2024. That version of Siri was shown handling context more naturally, understanding ongoing conversations, and taking actions across apps, but the feature has not yet been released to users, even after 1.5 years.

In a joint statement, Apple said it selected Google’s AI technology after “careful evaluation,” calling Gemini the “most capable foundation” for its plans. Apple also reiterated that Apple Intelligence will continue to run on-device or through its Private Cloud Compute infrastructure. While there is no clarity yet on whether Google will have access to Apple user data, the company stressed that its privacy standards will remain unchanged, suggesting Gemini models will operate within Apple-controlled systems rather than Google’s servers. A Bloomberg report from November had said Apple could end up paying around USD 1 billion per year to license Google’s foundational models.

Apple’s iPhones, despite being one of the most premium phones in the market, have lagged behind in terms of AI functionality. Analyst reports and social media feedback have repeatedly pointed out that Apple Intelligence is inferior when compared with AI features available on Android devices. Siri, in particular, has been criticised for limited understanding and slower progress compared to Google’s own assistant and other generative AI tools.

The deal also has implications for how Android and iOS differentiate on AI. Google has positioned Gemini as a core part of the Android experience, especially on Pixel phones, where AI features are a key selling point. By bringing Gemini to Siri, Apple narrows that intelligence gap. At the same time, it blurs the distinction between the two platforms, as both ecosystems may now rely on similar underlying models. The difference, going forward, is likely to come from how Apple and Google integrate AI into their respective user experiences rather than from the models themselves.

Apple has already been working with OpenAI, integrating ChatGPT into Siri in late 2024 for more complex queries that require broader world knowledge. That integration was widely seen as a temporary solution. Apple has told CNBC that there are no changes to its OpenAI agreement at this stage.

Google, meanwhile, has continued to advance its AI roadmap, introducing its upgraded Gemini 3 model late last year. Its Nano Banana image generation model is quite a hit amongst users.

For users, the partnership has finally raised expectations of a more capable Siri, but Apple has said it will not launch until 2026.