Sound, dynamic analysis of geosystems relies on accurate simulation of cyclic soil behaviour, for which numerous constitutive models with simple to sophisticated structures have been proposed over the past decades. Despite ongoing improvement reported from some advanced models in element test simulations and laboratory tests, the lack of examination in boundary value problems (BVPs) hampers their practical application. Recently, Cudmani et al. (first Int. Conf. on Geot. of Tailings and Mine Waste, 2023) proposed a set of numerical benchmarks of BVPs for validating and assessing the suitability of advanced constitutive models to capture the nonlinear soil dynamic response during earthquakes, including the phenomena of soil liquefaction. The numerical benchmarks include the shaking of a rigid block upon a 1) thin dry and 2) saturated soil layer, in-plane and anti-plane shaking of 3) homogeneous dry and 4) saturated soil layer, 5) horizontally layered soil deposit and 6) inclined homogeneous saturated soil layer. In this contribution, the aforementioned numerical benchmarks are selected and used to compare and assess the capabilities of the reference and newly proposed hypoplastic constitutive models to capture the nonlinear behaviour of coarse soils. Qualitative and quantitative comparisons are presented, and the similarities and differences in the hypoplastic models’ performances are discussed. The simulation results indicate plausible responses that the newly proposed models deliver, which are, in principle, able to capture the main features of soil behaviour during earthquakes. Nonetheless, high-quality experimental benchmarks, including advanced laboratory testing and physical models (e.g., 1 g and ng-shake table tests), are necessary to assess conclusively the predictive capability of the constitutive models.