Previous radiative parameterization evaluation of aerosol effects has been limited to participant model intercomparison on a handful of idealized cases. These few cases fail to capture the variety of atmospheric conditions and are unrepresentative of the model’s actual radiative transfer errors. This is especially problematic for understanding errors in the aerosol direct instantaneous radiative effect (IRE) that result from spatially varying optical properties which differ greatly between GCMs. We detail a new paradigm for benchmarking a GCMs radiation code. This technique will be used to evaluate participating GCMs in CMIP6 RFMIP Aerosol Instantaneous Radiative Forcing Protocol. It works by taking a snapshot of the atmospherics state and aerosol optical properties at every grid point and using this as input into a highly-accurate multi-stream doubling-and-adding radiative transfer solver. Thereby, creating a benchmark aerosol IRE that is unique to that GCM. The benchmarks for GFDL AM4 and CESM 1.2.2 show that characteristics of the IRE errors have a detailed spatial pattern that is unique to each model. The radiative parameterization error in aerosol IRE is up to 20% in the case of global average daily mean atmospheric absorption by AM4 and higher in localized regions.