Apple might use Gemini to power its revamped version of Siri, Bloomberg reports. The companies are in the early stages of exploring a partnership, with Google training a version of its Gemini model that can run on Apple’s servers. The iPhone maker was previously reported to be considering similar partnerships with OpenAI and Anthropic.

“The company is simultaneously developing two versions of the new Siri: one dubbed Linwood that is powered by its models and another code-named Glenwood that runs on outside technology, ” Bloomberg writes. A final decision has yet to be made on what AI model Apple will ultimately use, but it’s entirely possible it sticks with its internally-developed models.

While seceding a potentially key technology to a competitor seems decidedly un-Apple, it’s a testament to how much difficulty the company has had rolling out Apple Intelligence. Apple included the majority of its new AI features in the release of iOS 18, but failed to ship an updated version of Siri that can draw on personal data to take action in apps. Apple ultimately admitted the Siri update was delayed in March, with the new launch now expected sometime in 2026.

The turmoil caused by the delay reportedly led to Apple’s various AI projects changing hands internally, and the company’s decision to consider using a third-party AI model. As of the launch of the Pixel 10, Gemini now offers a similar set of capabilities to what Apple promised with the new Siri. It could presumably do the same things on iOS with the right changes. Using a version of Gemini that runs on Apple’s hardware and servers — the company calls this Private Cloud Compute — could also add an additional layer of security.

Apple was reportedly planning to offer Gemini as an alternative to ChatGPT in Apple Intelligence, but the option has yet to ship. Having Gemini power some of its AI features would put Apple in a similar position to Samsung, who’s Galaxy AI relies on a mix of custom models and Gemini.