It definitely answers certain questions better and languages, topics these models are less capable at are hit less. The denial at this point doesn't make any sense.