How about the NHC? (they combine them all for an average and are the most accurate imo)
Katrina:
Ps: I know thats not the question you asked but it the only model I put much stock in. Just showing you how accurate they were 3 days out on Katrina. ;)
The models had converged on that track about 16 hours before that point, IIRC.
The top overall model for Katrina was indeed GUNA, which is an average of the UKMET, GFS, GFDL, and NOGAPS.
The top individual model overall was the UKMET, though many were close behind.
The top performer for Katrina for 5 day forecasts was NOGAPS. The top performer for Katrina for 3 day forecasts was the BAMD and UKMET. (All the previous data actually automatically calculated by some software I have.)
The closest model to Katrina's actual landfall from precisely 3 days before landfall was the GUNA Consensus.
The current GUNA consensus for Rita is a landfall at East Matgorda Bay, a few miles east of the actual town of Matagorda, right over the little town of Gulf.