Accurate solar radiation data are required for solar energy development. In absence of long-term ground measurements, practitioners rely on modeled data, which typically is of unknown uncertainty. This article benchmarks solar irradiation estimation models over Uruguay, analyzing uncertainty behavior and providing recommendations. The performance of six models is evaluated using controlled-quality ground measurements for an extended period, being one of the few benchmark studies in Latin America. The LCIM, a model specially adapted for the region and based on GOES-East satellite images, exhibits the highest accuracy and spatial consistency with a remarkably low root mean squared deviation of 6% and mean bias of less than 1%. The NSRDB and GL1.2 estimates have also a competitive performance and are well-suited alternatives. The MERRA2 database presents high deviations and should not be an option for solar resource assessment in the region without post-processing. This research is a first step towards a South American benchmark and provides information on which estimation models are suitable for large-scale solar energy projects in the region.