Spatio-temporal features are considered of statistically significant stable change of surface air temperature obtained from 20th and 21st century simulations using a 17-member ensemble of global climate models forced with evolving green house gas and sulfate aerosol atmospheric concentrations. The analysis shows that the tropical Indian Ocean and western Pacific, on the one hand, and the northern North Atlantic and Southern Ocean, on the other hand, stand out as the most contrasting features of the ensemble mean timing distribution. The former are characterized by the earliest warming, while the latter demonstrate latest or not achieved stable changes of the temperature. Qualitative agreement between the model- and observationally derived global pictures of the stable change timing offers an opportunity to test model performance in reproducing the forced climate evolution. An illustration is given of the model performance evaluation based on the test application.