Stable Diffusion models to be hosted locally on Google Pixel 10 and Pixel 10 Pro with other AI features coming too
Kamila Wojciechowska claims to have discovered various hints about future AI features that Google plans to introduce across a host of Pixel devices in the next few years. For reference, Wojciechowska continues to cite a source from Google's gChips division for all of her information relating to unreleased Tensor chipsets.
Specifically, Wojciechowska speaks of Google actively developing Video Generative ML, which is expected to use AI algorithms in Google Photos and perhaps even YouTube Shorts to improve video editing. Moreover, Wojciechowska also mentions 'Speak to Tweak' and 'Sketch-to-Image', the latter of which is already available in one form or another across some recent Samsung Galaxy devices.
On top of that, a new 'NanoTPU' could debut with the Pixel 11 series to help monitor sleep tracking, fall detection and sleep apnea. Furthermore, Stable Diffusion-based models are expected to run locally from the Pixel 10 onwards rather than the cloud-based solution that Google currently offers.
Allegedly, the new image signal processors (ISPs) and Tensor Processing Units (TPUs) in the Tensor G5 and Tensor G6 are the source of these AI-led improvements. Reportedly, Google does not plan to introduce any of these new features until the Pixel 10 at the earliest, which should land in summer 2025. However, there is every chance that these will be delayed a year to coincide with the release of Android 17 and possibly the Pixel 11 series.
Source(s)
