The above animation shows decadel averaged temperatures for Norh America as calculated by the GISS-E2-R1_r1i1p2 model for RCP8.5 emissions scenario. GISS-E2 is my favoured model because it best reproduces a fit I made a year ago to HADCRUT4 (see below). The next plot shows 31 CMIP5 models compared to HADCRUT4 (black points), the fit shown in red and GISS-E2-R in blue.
The original fit was made to a logarithmic temperature dependence on CO2 (Mauna Loa) superimposed with a 60y natural oscillation. The details are given below.