Regarding the ensemble intensities, those are taken straight from the GFS ensemble raw wind fields. The ensemble members are all run at a relatively coarse resolution, about 100km, which does not allow them to represent tropical cyclones very well. While they can and often do have a decent initial intensity -- that they get from the main GFS analysis as a first guess -- they often can't sustain it at later times. It's mostly included in the suite of plots for completeness; the CSU page does not output those intensity fields.

OFCI is the 6hr interpolated NHC official forecast, meaning it is from the previous forecast cycle but interpolated 6 hours ahead in time. I see that the 18z intensity guidance plot had the storm going down to 55kt. The only reason it does so is because the 12z package put out by the NHC (the 11a ET advisory, basically), went from 120kt to 80kt in a 24-hr period at the end of the 5-day forecast period. Interpolated ahead another six hours, that leads to further weakening of the storm (though 55kt is overkill). As with any of the various model pages out there, the raw values should be taken with caution.

(Note that the CSU plot at the same time had the same OFCI intensity.)

Hope this helps and isn't too technical...