In a recent episode of the “Made by Google” podcast, Siyang Xiao, Vice President of Hardware and Services Software, stated that the Pixel 8 Pro, equipped with 12 GB of RAM, was the perfect platform to download the “Gemini Nano” model and explore the available capabilities. However, the base version of the Pixel 8 comes with 4 GB less RAM, and it seems that Google did not want to “compromise the quality of the experience” by limiting the AI model to only 8 GB of RAM.
Google has recently had a change of heart – somewhat. In a recent announcement on the Pixel phone support site, the tech giant mentioned that it will introduce the Gemini Nano option for Pixel 8 users as an available choice for developers in the upcoming software update. This is good news for developers who know how to enable it, but for most regular Pixel 8 users, it is not a hugely significant development – at least not at the moment.
Returning to Xiao’s appearance on the podcast, the Google employee explained that the company wants some AI features, like smart replies, to be “resident in RAM”, meaning they will be running “permanently” in the memory, allowing them to be ready to work at any moment.
It is worth noting that Google informed Ars Technica that both the Pixel 8 and Pixel 8 Pro phones do not currently retain the Gemini system in memory, and users need to activate the developer option to do so.
In short, Google released the Gemini AI model on some selected mobile devices earlier this year through an Android app, but some were surprised that one of the company’s latest flagship phones was not included. At the Mobile World Congress, it was revealed that the Gemini Nano model could not run on the Google Pixel 8 due to unspecified hardware restrictions. Now, we have a better idea of what was hindering the phone from being compatible with the AI within the device.
Armed with this new information, some may question whether they truly want to run the Gemini Nano app on their phones or not. Installing it without restrictions will result in a “permanent loss” of a portion of the system’s memory, potentially affecting the performance of other applications and services on the smartphone. Is this trade-off worthwhile?
