Damejune, look for general trends within the models, not just at the most current graphics. Keep in mind that being the most up to date solution does not necessarily imply the most accuracy. The GFDL, for example, has been somewhat consistent with the last three runs, oscillating more with intensity and speed than location, I think. Check out the 06z GFS, which took it through the Yucatan straight and landfalls somewhere north of Sarasota. Granted, it does like to nudge the storm west more and more with each run.
The cyclone’s eventual landfall is way out of the good timeframe for these things anyway , so there really isn’t much to say at this juncture.

This is going to be a painful wait, methinks.

EDIT: To answer your below question, you can do this in the ‘initial time column’. Scroll down and pick whatever one you want; 2005101612, 06, 00, etc (they do a new one every six hours).

Be careful with the gfdl, which is a limited area baroclinic model. To get the super detailed images of the hurricane's eyewall and crap like that, they have to focus on a small area, as it would be too time-consuming to render the entire basin. The different grids of focus are designated by storm name, so be sure to pick td24, not the east pacific thing.