Apple wants to integrate Google Gemini into Siri

Apple is considering integrating Google’s Gemini artificial intelligence into a new version of its Siri voice assistant, Bloomberg reports. The company is in preliminary talks with Alphabet to test how a third-party model could run on Apple’s servers and become the “brain” of the new Siri in 2025.

According to sources, the discussions are at an early stage. Google has already begun training a model that can work with Apple’s infrastructure. Apple has also previously considered collaborating with Anthropic and OpenAI to determine which of the models — Claude or ChatGPT — would be better suited for the updated assistant.

The reason for looking for an external solution was Apple’s lag in generative AI and delays in the long-awaited Siri update, which was supposed to be released last spring. The update was postponed due to technical problems, which led to the removal of John Giannandrea, who previously oversaw the project. Now Apple’s management, led by Craig Federighi and Mike Rockwell, is considering the possibility of attracting external technologies as one of the development paths.

Apple is simultaneously developing two versions of the new Siri: an internal model codenamed Linwood and an external one codenamed Glenwood. In the case of using Google Gemini, the model will run on Apple Private Cloud servers, and not on the devices themselves.

The company’s management is reported to decide on a partnership with Google or continuing to work with internal models in a few weeks. Representatives of Apple and Google declined to comment.