In recent years various researchers have proposed an optimal control approach for the rejection of turbulence-induced wavefront distortions in an AO system. The essential element in the design of an optimal controller is the choice for the turbulence model, which predicts the turbulence to compensate for the inherent delay in the AO control loop. In this paper various models as proposed in literature are considered; ranging from first order temporal models to high-order full spatialtemporal models. The various models are analyzed and the resulting 1-step ahead predictors are derived. The performance of the predictors are compared for a von Karman type of turbulence with frozen flow propagation in and time-varying propagation directions.