Recent observations have shown that the environmental quenching of galaxies at z & SIM; 1 is qualitatively different to that in the local Universe. However, the physical origin of these differences has not yet been elucidated. In addition, while low-redshift comparisons between observed environmental trends and the predictions of cosmological hydrodynamical simulations are now routine, there have been relatively few comparisons at higher redshifts to date. Here we confront three state-of-the-art suites of simulations (BAHAMAS+MACSIS, EAGLE+Hydrangea, IllustrisTNG) with state-of-the-art observations of the field and cluster environments from the COSMOS/UltraVISTA and GOGREEN surveys, respectively, at z & SIM; 1 to assess the realism of the simulations and gain insight into the evolution of environmental quenching. We show that while the simulations generally reproduce the stellar content and the stellar mass functions of quiescent and star-forming galaxies in the field, all the simulations struggle to capture the observed quenching of satellites in the cluster environment, in that they are overly efficient at quenching low-mass satellites. Furthermore, two of the suites do not sufficiently quench the highest mass galaxies in clusters, perhaps a result of insufficient feedback from AGN. The origin of the discrepancy at low stellar masses (M-* less than or similar to 10(10) M-circle dot), which is present in all the simulations in spite of large differences in resolution, feedback implementations, and hydrodynamical solvers, is unclear. The next generation of simulations, which will push to significantly higher resolution and also include explicit modelling of the cold interstellar medium, may help us to shed light on the low-mass tension.