Traffic grooming in optical networks employing wavelength division multiplexing (WDM) has gained prominence due to the prevailing disparity between the user requirement and wavelength capacity. Nodes in an optical network get upgraded to the latest grooming technology slowly with time. Hence, WDM grooming networks are expected to employ heterogeneous grooming architectures. In this paper, we develop an analytical model to evaluate the blocking performance of WDM grooming networks with heterogeneous grooming capabilities. We demonstrate the accuracy of the analytical model by comparing the analytical results with that of the simulation. We observe that analytical models with and without precise knowledge of the grooming architectures predict similar performance. The proposed analytical model can be employed by resource placement algorithms that identify a set of nodes and links that need to be upgraded when the resources are limited.