Real-time hybrid simulation (RTHS) is an experimental testing technique widely used for performance evaluation of structural systems such as large buildings and bridges subjected to earthquake loading. While RTHS testing has demonstrated over the last 20 years to be an efficient and cost-effective alternative to shaking table tests, especially for large structural systems with rate-dependent behavior, accurate and stable results from this methodology are highly dependent on the test specimen, loading equipment, and controller design for dynamic compensation. This paper presents a study on the accuracy and stability of model-based compensation (MBC) approaches for the implementation of a real-time hybrid simulation benchmark problem. The controller architecture is based on feedforward compensator, designed for reference tracking, while a feedback regulator provides improved robustness for undesired disturbance and sensor noise. The results provide evidence of the improved performance of MBC controllers compared to benchmark results. Moreover, the MBC controllers surpass the benchmark controller in terms of robustness, when multiple partitioning cases and control plant uncertainty are considered in the numerical simulations.