For PMV, Adaptive and UTCI, I would say that LBT is better. It has more ways to customize these models and it also has a better implementation of the SolarCal model that accounts for shortwave solar irradiance falling onto occupants (for the MRT input of these models)
For thermal comfort mapping, it’s still a bit of a wash between Legacy and LBT until I can finish implementing a robust longwave MRT calculation in LBT. This should happen soon, at which point LBT thermal comfort mapping will be significantly better than Legacy.
For PET, we haven’t implemented it yet in LBT so Legacy is clearly better for that model.
For most of the other comfort models that you would get with the Legacy “Thermal Indices” component, they’re implemented in the ladybug_comfort SDK so you can make use of them with a line or two of Python in a GHPython component. I just haven’t figured out yet how I’d like to expose these models on Grasshopper components or if it’s better to just leave them on the SDK layer given that they are not as popular.