Google has more AI tools than most of us can keep up with so it is possible (likely) that you haven’t heard of AI Edge Gallery.
It gives you an easy way to experiment with small language models on your phone or tablet (it technically also works on laptops but doesn’t feel like it was designed for them).
You can download the app, install whichever models sound interesting (like Gemma 4), try them out, then either keep them forever (unlikely with how much things are changing) or remove them just as easily as you installed them.
One of the big selling points of something like this is that these models do not go back to cloud servers. Anything that you have them create stays local to your device. This has privacy, security, and availability implications.
The smaller models are an “acquired taste” and may not be for people who are new to the AI space. I do think that this will eventually be commonplace so that people can just have small models that can handle the exact tasks that they care about.
Time will tell I suppose.
If you would like more information, here is the site - https://developers.google.com/edge/gallery