The thermal performance of adaptive building envelopes can be evaluated using building performance simulation tools. Simulation capabilities and accuracy in predicting the dynamic behaviour of adaptive building envelopes can be enhanced through co-simulation. However, it is unclear how accurately co-simulation can predict the performance of adaptive building envelopes and how the accuracy of adaptive building envelope models created in co-simulation setups can be assessed and validated. Therefore, this study presents new evidence on the empirical validation of co-simulation setups for adaptive building envelopes by establishing an assessment framework to determine the extent to which they can accurately represent the real world. The framework was applied to a case study to validate a co-simulation setup for a blind automation system using monitored data from MATELab, a full-scale outdoor test facility with realistic indoor and outdoor conditions. The validation of the co-simulation model of MATELab resulted in a median CV-RMSE index, a measure of model accuracy, of 5.9%. This indicates that the simulated data points have a small variance relative to the measured data points, showing a good model fit. In the future, modellers from the façade community can use the assessment framework for their co-simulation setups.