I believe with the better data input the track won't very too much from what it is now, maybe a bit west, maybe a bit east, but generally what we see is what we get, including the possibility for further up the east coast and out to sea, or slightly west for an north central or eastern Gulf hit. Middle is Florida, and this has shifted from NC/SC to here to be the most likley. (Although there is still a great deal of uncertainty) I think the models may trend east a bit, then back west over the next few days, it's going to be close enough to cause watches/warnings along Florida I do believe, maybe as early as Friday afternoon/Evening for eventual landfall somewhere Sunday or Monday.

HMON (replacement for the older GFDL) is a hurricane specific model that tends to do pretty well with developed systems like Irma, it only goes out a few days as well. The models in general are pretty tightly in agreement in the 4 day range. HWRF is similar, both show Irma become very large and intense systems, the ocean temps north of Cuba are some of the highest in the Atlantic so the energy is there.

HMON run keeps it north of Cuba and gets below 900mb, HWRF passes it right through Cuba (ends with it barely back in the Caribbean then turning it right back north into Cuba 955mb vs 872mb on the HMON, which doesn't take it over land)

Irma is still east of the Leewards, so we have a few days to watch it.