Apple in Talks with Google to Develop New AI Gemini-Powered Siri

Apple is in talks with Google regarding the use of Google's AI Gemini as the foundation for the next version of Siri. According to Bloomberg, the discussions are in the early stages, but Apple has already approached Google with a request to create a custom AI model that will be used for Siri. Google has begun training such a model, which will operate on Apple Cloud Compute's private servers.

Previously, Apple had negotiations with OpenAI and Anthropic, and both companies are working on versions of their models for Apple's servers. While Apple is exploring partnerships with various AI companies to enhance Siri, a final decision on using third-party AI solutions or developing its own models has not yet been made. The company is currently testing several LLMs, including its own, to determine the most effective option for users.

Two versions of the new Siri are currently being developed: one based on Apple's models and the other on third-party solutions. Apple also aims to create a more advanced version of Siri that will completely transform the assistant's architecture. The company's chief engineer, Craig Federighi, referred to this new architecture as the "second generation" of Siri. He noted that the "global modernization" has enabled the achievement of results necessary for more complex AI functions, such as personalization, which were promised under Apple Intelligence in iOS 18.

Federighi added that this modernization allows not only for the completion of planned tasks but also provides a significant upgrade to Siri's capabilities. Although Apple considered collaborating with Anthropic, changes in the company's structure forced it to reassess its plans and consider other AI companies. As of now, no agreement has been reached with Google, OpenAI, or Anthropic. Rumors suggest that the launch of the new Siri based on LLM could be expected in spring 2026.