The first step in building a generative model of molecular dynamics that can be used to accurately infer intrinsically disordered protein (IDP) kinetics is to evaluate the current SoTA model. We will run genAI MD simulations of a minimal set of IDP sequences using a published implicit transfer operator model equipped with protein language model embeddings that reaches SoTA performance on equilibrium sampling tasks for structured proteins. We will then compare the generated ensembles to a set of long-timescale classical MD simulations performed on the same systems and experimental observables. This work will allow us to identify the efficacy of the current SoTA model and determine failure modes to motivate future improvements.