Land surface skin temperature, a critical indicator of climate change, connects the water and energy cycles between the land and the atmosphere. Here, we evaluate the simulations of land surface skin temperature from the Coupled Model Intercomparison Project Phase 5 (CMIP5) and CMIP6 models with satellite‐based datasets and reanalysis. We find systematic cold skin temperature biases over arid regions in CMIP5/CMIP6 simulations. Over arid and semi‐arid regions, latent heat biases drive skin temperature biases by evaporative cooling. Over humid regions, surface downward shortwave and albedo biases are relatively more critical. Spatial patterns of biases remain similar in the latest CMIP6 simulations, suggesting systematic biases in land‐atmosphere interactions. These biases need to be corrected or considered while using models for future projections. Plain Language Summary Earth system models are used for future climate assessment. This study diagnoses the model performance of variables in representing the land surface energy budgets. Results show systematic and persistent cold biases over arid regions in the latest Coupled Model Intercomparison Project Phase 6 simulations, associated with overestimated evapotranspiration cooling. The aridity‐dependent biases in skin temperatures show modeled deficiencies in land‐atmosphere interactions. Results suggest a higher priority to develop and improve related processes to eliminate temperature biases in state‐of‐art climate models in dry conditions. Key Points Systematic wide‐ranging cold biases are found in the modeled land surface skin temperature over arid regions Spatial patterns of biases remain similar in Coupled Model Intercomparison Project Phase 5 (CMIP5) and CMIP6 simulations These temperature biases must be corrected before the models are used for future projections