Understanding the conformational behavior of proteins has been a long-standing challenge to which computer simulations could contribute significantly. This concerns the folding behavior of proteins as well as the conformational statistics of intrinsically disordered proteins. A well-converged sampling of the conformational statistics over a broad range of control parameters is computationally extremely challenging and best addressed by coarse-grained modeling, for example, with an intermediate resolution model like the PRIME20 model. A comprehensive understanding of the thermodynamics and conformational statistics of individual protein chains is, however, not only a goal in itself but also a prerequisite for a better understanding of their aggregation tendency into oligomers and further into amyloid fibrils. We present here an extensive comparison of the ability of the PRIME20 model (in its documented variants in the literature) to faithfully reproduce the thermodynamics and statistics of three homopeptides having very different folding temperatures and different typical secondary structures.