I don't think there is a problem with discussing long-range model output for existing storms here, but it is important to keep in mind that model runs get stale pretty quickly... old runs don't immediately become worthless whenever a new run comes in, but by the time they are 2-3 days old, they pretty much are worthless, except in a retrospective sense to determine model skill.

Regarding Dean, there is still a lot of variability in the forecast tracks, even in the short and medium range. The 12Z GFS is a lot further south with the track and the 12Z GFDL is much further south and much weaker with the system at 126 hrs compared to recent runs.