Climate models exhibit a broad range in the simulated properties of the global climate. In the early historical period, the absolute global mean surface air temperature of models contributing to the fifth phase of the Coupled Model Intercomparison Project (CMIP5) spans a range of ~12-15 °C. Other climate parameters are linked to the global mean temperature, such as sea ice area, atmospheric circulation patterns, and by extension cloudiness, precipitation and albedo. Accurate representation of the baseline climate state is crucial for meaningful future climate projections, since the baseline conditions may dictate the capacity for change. For example, a model with initially smaller sea ice area has less potential to lose sea ice as the planet warms. Amongst the CMIP5 models, it is found that in the baseline climate state there are coherences between Southern Ocean temperature, outgoing shortwave radiation, cloudiness, the position of the mid-latitude eddy-driven jet, and Antarctic sea ice area. The baseline temperature relationship extends to projected future changes in the same set of variables. The tendency for models with initially cooler Southern Ocean surface temperature to exhibit more global warming, and vice versa for initially warmer models, can therefore be linked to baseline Southern Ocean climate system biases. A first look at emerging data from CMIP6 reveals a shift of the tendency towards the Antarctic region, potentially linked to a reduction in biases over the Southern Ocean, which prompts an examination of biases in the Antarctic region as more CMIP6 model data becomes available.