Since 0.1 nm is comparable to the size of an atom, it means that on
average all atoms are off by one atom-atom distance. Some regions will
usually be better (helix, sheet), others worse (loops). At the
percentage of sequence identity between target and template (about
35%) this is about the accuracy that one can expect, and for many
purposes (functional studies, drug design, etc) the model structure
would be a valuable starting point if no experimental structure is available.