Is this just a terminology thing? KiCad does not use the term “antipad”, instead it calls this “via diameter” or “clearance” depending on what you mean by it
The problem with using via clearance is that when using a padstack with different annular rings on some layers, using clearance yields different effective clearance diameters per layer when per antipad definition one wants this value to be constant through the entire padstack
The last answer by @JeffYoung right on addresses the antipad problem:
Gitlab issue linked above: #19588 (closed)