Pixel 9 series will debut a more powerful version of Gemini Nano

Google I/O has officially kicked off with a ton of new announcements about its current and future AI plans. A key reveal from the keynote focused on Google’s imminent flagship range, the Pixel 9 series, and its new AI capabilities. The company has confirmed that the upcoming Pixels will be the first to sport an upgraded version of Gemini Nano with multimodal capabilities.

That means the on-device AI model on the phones will not just be capable of processing text inputs but will also be able to discern context from real-world images, sounds, and spoken language. Google recently posted a teaser of how this might work. The video shows a demo of an AI explaining the scene in front of it and having a conversation with the user about it.

Gemini Nano currently powers limited features on the Pixel 8 lineup, including the new Pixel 8a. Specifically, it runs Summarize in the Recorder app and Smart Reply in Gboard. The new Gemini Nano version coming to “Pixel later this year” will be much more useful than the current version.

Google says the improved Gemini Nano will also run a new feature in TalkBack that will help people with blindness or low vision describe what’s happening in an image more clearly. “Whether it’s more details about what’s in a photo that family or friends sent or the style and cut of clothes when shopping online,” the feature will be able to generate on-device descriptions of unlabeled photos without the need for a network connection. You can see a preview of this feature in the video above.

Will the new Gemini Nano come to older Pixel phones?

It’s unclear if Google plans to bring Gemini Nano’s multimodal capabilities to existing Pixel phones. During a press briefing, the company clearly stated that the upgraded AI model needs new Neural Processing Units (NPUs) and quite a bit of RAM and that it will be available “starting with the new Pixels later this year,” followed by other devices launching in that timeframe.

By the sound of it, the new Gemini Nano will require additional, more powerful hardware and is unlikely to make it to the Pixel 8 and older models unless Google comes up with a workaround. Google did say that it’s working on making its AI models more compact so that they can achieve on-device on a wider range of devices. The company will share a complete list of supported devices for the more powerful Gemini Nano later this year.