Large-scale experiments show that state-of-the-art dataset distillation methods are comparable to or worse than coreset selection on ImageNet and ImageNette. Coresets consistently achieve better data coverage and are more computationally efficient, highlighting their practical superiority over distilled sets.