Because the forcings to which Coupled Model Intercomparison Project - Phase 5 (CMIP5) models were subjected were poorly quantified, recent efforts from the Radiative Forcing Model Intercomparison Project (RFMIP) have focused on developing and testing models with exacting benchmarks. Here, we focus on aerosol forcing to understand if for a given distribution of aerosols, participating models are producing a radiometrically-accurate forcing. We apply the RFMIP experimental protocol for assessing flux biases in aerosol instantaneous radiative effect (IRE) on two participating models, GFDL AM4 and CESM 1.2.2. The latter model contains the RRTMG radiation code which is widely used among CMIP6 GCM's. We conduct a series of calculations that test different potential sources of error in these models relative to line-by-line benchmarks. We find two primary sources of error: two-stream solution methods and the techniques to resolve spectral dependencies of absorption and scattering across the solar spectrum. The former is the dominant source of error for both models but the latter is more significant as a contributing factor for CESM 1.2.2. Either source of error can be addressed in future model development, and these results both demonstrate how the RFMIP protocol can help determine the origins of parameterized errors relative to their equivalent benchmark calculations for participating models, as well as highlight a viable path towards a more rigorous quantification and control of forcings for future CMIP exercises.