It reads how your JPG was saved
A JPG has already been compressed once, and how much more it can lose depends on the quality it was saved at. ToolGPT reads that quality from the file itself before it estimates anything. With that number, the size on each option is a ceiling: the new file will be no larger, and it usually comes out smaller.
Because of this, an option that cannot help is not shown. If a JPG was already saved at about the quality the High quality option uses, re-encoding it would cost sharpness and add bytes, so that option is left out. If no option can make the photo smaller, ToolGPT tells you why, and a smaller pixel size is usually the way forward.