Both local luminance as well as visual context affect perceived reflectance (lightness) and perceived luminance (brightness). Various computational models try to capture mechanisms underlying these percepts, mostly for simplified stimuli where lightness and brightness cannot be distinguished. While each of the proposed models captures some lightness and brightness effects better than others, a comparative evaluation of their overall performance is difficult for the following reasons: (1) no commonly accepted set of stimuli serves as benchmark to test these models on; (2) openly available model implementations lack a consistent interface, and/or rely on closed-source software; (3) models without available implementations are difficult to re-implement due to ambiguous descriptions of stimulus- and model-related parameters. To address these problems we present an open-source framework for testing and comparing models of human brightness perception in Python. For this, we provide implementations of many commonly used stimuli and tools for producing parametric variations of these. We also provide implementations for various existing models of brightness perception including several multiscale spatial filtering models and an edge-based model. Finally, we provide example workflows for evaluating these models on sets of stimuli, both replicating published results as well as challenging models with novel stimuli and model parameterizations. This platform provides a better overview of the advantages and limitations of existing and future models of human brightness perception. Additionally, we hope to start a discussion about stimuli appropriate for benchmarking models and about standardization of interfaces for computational models. [Funded by the Deutsche Forschungsgemeinschaft (DFG, German Research Foundation) under Germany's Excellence Strategy – EXC 2002/1 "Science of Intelligence" – project number 390523135.]