[Pharo-users] Why is there both a standard VM and a headless VM?

This may be a silly question but why are there two VMs? Since the headless VM seems to do everything the standard VM does plus you can write proper Gtk apps with it, why would you ever use anything else? I seem to recall someone saying a while ago that the headless VM had not been optimised yet so was slower than the standard VM. Is that all it is?