The Atlantic Subtropical Cells (STCs) consist of poleward Ekman transport in the surface layer, subduction in the subtropics, and equatorward transport in the thermocline layer that largely compensates the surface Ekman divergence and closes the STCs via equatorial upwelling. As a result, the STCs play an important role in connecting the tropical and subtropical Atlantic Ocean, in terms of heat, freshwater, oxygen, and nutrients exchange. However, their representation in state‐of‐the‐art coupled models has not been systematically evaluated. In this study, we investigate the performance of the Coupled Model Intercomparison Project Phase 6 climate models in simulating the Atlantic STCs. Comparing model results with observations, we first present the simulated mean state with respect to ensembles of the key components participating in the STC loop, that is, the meridional Ekman and geostrophic flow across 10°N and 10°S, and the Equatorial Undercurrent (EUC) at 23°W. We find that the model ensemble reveals biases toward weak Southern Hemisphere Ekman transport and interior geostrophic transports, as well as a weak EUC. We then investigate the large inter‐model spread of these key components and find that models with strong Ekman divergence between 10°N and 10°S tend to have strong mixed layer and thermocline interior convergence and strong EUC. The inter‐model spread of the EUC strength is primarily associated with the intensity of the southeasterly trade winds in the models. Since the trade‐wind‐induced poleward Ekman transports are regarded as the drivers of the STCs, our results highlight the necessity to improve skills of coupled models to simulate the Southern Hemisphere atmospheric forcing. Plain Language Summary This work systematically assesses how well state‐of‐the‐art climate models simulate the Atlantic Subtropical Cells (STCs). The STCs are part of the shallow (upper 200 m) ocean circulation. They connect the upper tropical and subtropical Atlantic Ocean in each hemisphere via poleward transport in the surface layer (surface to about 100 m) and equatorward transport in the subsurface layer (100–200 m). Therefore, the STCs are important for the exchange of heat, freshwater, dissolved oxygen, and nutrients between these regions. In this work, we find that the climate models generally simulate a too weak Southern Hemisphere STC which can be related to weaker southeasterly trade winds compared to observations. The results highlight the necessity to improve the ability of coupled climate models to realistically simulate the winds in the Southern Hemisphere, in order to better simulate the tropical‐subtropical ocean circulation. The results provide an important reference for studies on the mean state and variability of the Atlantic STCs, and for studies investigating the STCs' possible change in the future under global warming using coupled climate models. Key Points The performance of the Coupled Model Intercomparison Project Phase 6 climate models in simulating the mean Atlantic Subtropical Cells is evaluated The model ensemble shows large biases toward weak southern hemisphere subtropical cell and weak Equatorial Undercurrent The large inter‐model spread of the subtropical cell and Equatorial Undercurrent strength is associated with the southeasterly trade winds