Apple will reportedly use a custom version of Gemini to power the new Siri

When Apple introduces its more capable version of Siri in 2026, it increasingly sounds like it’ll have Google to thank. Bloomberg reports that Apple is finalizing plans to use a custom version of Gemini to power the new Siri and plans to pay Google handsomely for the privilege.

The new Siri will still use some of Apple’s homegrown models, but will also rely on a version of Gemini running on the company’s Private Cloud Compute servers to “handle Siri’s summarizer and planner functions,” Bloomberg writes. Those functions “help the voice assistant synthesize information and decide how to execute complex tasks.” Since using apps on people’s behalf is core to Apple’s pitch for its updated assistant, Gemini appears to be playing a critical role.

Apple will reportedly pay $1 billion annually to use Google’s technology, a fraction of what Google has reportedly paid the company to make Google Search the default search option on Apple’s devices. The partnership will likely not be advertised, though, and Apple plans to eventually replace Google’s model with its own. Specifically, with “a one trillion parameter cloud-based model that it hopes to have ready for consumer applications as early as next year,” according to Bloomberg.

Siri can use ChatGPT for specific requests as part of Apple Intelligence, and Gemini was rumored to be added as another AI option in March 2024. After Apple delayed Siri a year later in 2025, though, the need for more substantial help reportedly became necessary. The possibility of Siri being backed by Gemini was first reported in August, when Google was asked to create a version of its model that could run securely on Apple’s servers. The iPhone maker was previously rumored to have explored deals with both Anthropic and OpenAI before that.