Nonsense. Try plotting the HadCRUT4 you posted or since you want the more up to data the HadCRUT4.4.0 data on the graph that you originally provided.
Oh, I forgot, you don't understand how to.
Fortunately, Gavin Schmidt has already plotted the current data against the current CMIP5 run of the models.
https://pbs.twimg.com/media/CIRz_2PVEAAf8QX.jpg:large
His graph confirms the observed data are nowhere near the average of the CMIP5 models. Confirming -- yet again -- that the predictions have been spectacularly wrong.

