It is worth noting that the official forecast track has an average 5-day error of 429 miles with Jeanne. The AVN has been the best performer, with an average 5-day error of just 150 miles. None of the other models I'm looking at have come close; next closest is the 5-day climatology & persistence (CLIPER) model at 261 miles at 5 days, which shows you how bad things have been.

Nonetheless, I've attached a plot I created with StormTrakker 6 (beta) depicting all available model runs of Jeanne. I've noted a few of the big ones on the map itself. There is a small cluster of models that turns the system off of the Florida coast, a much larger grouping that turns the system very near the Florida coast or over the state itself, and another small grouping that takes the storm across the state into the Gulf.

Of note is the AVN path, which is a hybrid of Frances' until you get to Osceola County and Charley's thereafter through Orlando and out to sea again, as well as the GFDL path, which skirts almost the Space & First Coasts. The BAM-series consensus is also noted, which probably isn't needed, but shows a path across the state.

Based upon the available guidance, I'd expect the track to be shifted a bit further to the left at 11pm. Note that all of these model runs are from 9/22 12z, 9/22 18z, or 9/23 00z, depending upon the model.