That's because git-grep doesn't need to bother to go through the filesystem for the on-disk files. It greps directly from the object storage instead. And AFAIK

That's because git-grep doesn't need to bother to go through the filesystem for the on-disk files. It greps directly from the object storage instead. And AFAIK it can run the grep in parallel. Practically any non-archaic machine has multiple cores now, so git-grep can easily be faster than regular grep from command line.