I'm really not sure why you guys picked the tasser1 model over the Zhang4 model. Zhang4 has more contacts made and a less "messy" structure overall. Oh well.
Might have fewer contacts in Foldit (don't know!), but that is well overshadowed by the gorgeous structure Zhang4 has.
By our methods, the TASSER model appeared to fit this set of contacts slightly better than the Zhang model. However, I agree that the Zhang model looks much more realistic. It might be interesting to run another puzzle with the Zhang model.
Baker 5 and Quark 4 also produced high contact scores which competed with Zhang for realistic modeling (in my opinion, I am no scientist [yet]). Would love to see all of them available for fixing up :)