Well I filmed it, so it's spot on! Though I would say that.
The physics models are always a little bit generous - because I'm testing my own rides I can tell you I did a 7:41 Cootha Back virtually with 329W a couple of weeks after I filmed it at 8:21 with 325W. Mind you my bike weight is very much 'ideal' in Fulgaz (no water, tools, clothing helmet etc.) though I suspect IRL rolling resistance is a bit worse.
Gravatt I haven't tested at full tilt yet but I filmed it at 6:34/303W and rode it virtually while tapering at 7:07/238W so it's probably about right. The gradients feel correct. I've just given it final approval so it should be up there in a week or two.
Inevitably the 'holder' of the virtual KOM did it over a minute faster than the IRL KOM. As part of a 'bike ride' of 605km at an average 6.1km/hr. There seems to be a move on Strava to move a ride to 'virtual' if it gets flagged as a 'ride'. Idiot.