The few things a model does better than code — and only those.
A model earns its place here by doing something code cannot. Reading handwriting off a photograph is one of those things — no amount of clever image processing gets there, and it is why this category exists at all.
It is deliberately a short list. Most of what gets labelled AI is a text box in front of somebody else's API, and a tool that could have been fifty lines of JavaScript is worse for being a model call: slower, less private, and it costs someone money every time you press the button. Everything else on Pathine runs on your own device precisely because it can.
These do not all work the same way, and the difference is worth knowing. Removing a background and enlarging an image run their models inside this tab — your picture never leaves the device, and there is no limit on how often you use them. Reading text, describing an image, summarising and rewriting need models far too large to download, so those send your file to a server.
Every page says which it is at the top, before you hand it anything. The ones that use a server hold your file in memory for the second or two the request takes, never write it to disk, and have a daily allowance — the honest shape of a free tool that costs real money to run.