dem05: The GFS isn't going to capture the intensity or structure of a tropical system very well, so it's not really to be trusted for tropical development. What you see therein the model is about as close as you are going to get. Case in point: Isabel at its peak intensity was no lower than 996mb in the GFS. The track is okay, as the influences that influence track from the midlatitudes are fairly well represented (but the weaker intensity of the storm negatively impacts the track forecast nonetheless), but I wouldn't put too much stock in the intensity of the storm in the model.