Actually, the models are unusually close together for a storm this far out. Much less unpredictability than usual.
And there's a real problem with people who don't really know anything about any of the models simply looking at maps with colored lines.
The only models pointing at Florida are the completely useless ones.
BAMM is the Beta-and-Advection Model, MEDIUM Layer. It's designed for BADLY SHEARED weak tropical storms. It's useless for Hurricanes. It only shows the low and medium level steering flow.
And the A98E is the oldest and most useless model that is still run. It incorporates a lot of climo (based on history of previous storms, rather than current weather conditions.
In Andrew, the Climo models showed Andrew curving harmlessly out to sea. The Dynamical models showed Andrew hitting Florida five days before it did.
As Avila noted in the latest discussion, there are no global dynamical models getting Isabel anywhere near Florida.