Extracting regional precipitation predictability on sub-seasonal to seasonal timescales from low frequency variability phenomena, like El Nino Southern Oscillation (ENSO), remains a challenge. Remote teleconnections to ENSO are regulated by the Rossby waves emanating from the Tropical Pacific that are excited by deep convective activity associated with SST anomalies there. Sub-grid scale deep convection parametrizations, however, are one of the largest sources of error in the current generation of earth system models and thus also lead to erroneous simulations of ENSO teleconnections. Here, we evaluate the simulation of ENSO teleconnections to US winter precipitation mean and extremes in a hierarchy of earth system models spanning spatial resolutions from 100 km to 25 km as they resolve finer scales. We also evaluate US Department of Energy’s Energy Exascale Earth System Model’s Multi-scale Modeling Framework (E3SM-MMF) model configuration that embeds a cloud resolving model in each model grid point to explicitly resolve deep convection. We find that models credibly simulate the observed ENSO teleconnections to US winter precipitation mean and extremes over Southeast and Southwest US - we also account for ENSO diversity. High resolution models and the MMF model are generally found to improve upon their low-resolution counterparts. The enhancements are largely due to the improved simulation of ENSO-dependent storm track activity and moisture fluxes into the regions.