An appropriate same-game test would probably pit them each against an assortment of monsters that tend towards physical combat (as opposed to being spellcasters in disguise). Say, a Giant, a Dragon, a Devil/Demon, and something else, all of appropriate CR to the level being tested.
@Level 5: Hill Giant (CR7), Hellcat (CR7), Juvenile White Dragon (CR6), Seven-Headed Hydra (CR6)
@Level 10: Fire Giant (CR10), Hezrou (CR10), Young Adult Black Dragon in a swamp (CR9+1), Dread Wraith (CR11)
@Level 15: Storm Giant Fighter 2 (CR15), Nalfeshee (CR14), Mature Adult Blue Dragon (CR16), Greater Stone Golem (CR16)
@Level 20: Balor (CR20), Old Red Dragon (CR20), Nightcrawler (CR18), Tarrasque (CR20)