Aquaro et al. (2021) consider a heterogeneous version of the standard spatial autoregressive (SAR) panel data model whereby all coefficients, including the spatial lag coefficient, are allowed to differ over the cross-section units. They refer to this more general specification as the heterogeneous SAR (or HSAR) model. The latter is the product of convergence of two very active streams of the recent panel data literature: (homogeneous) spatial panel data models. In this paper we describe hetsar, a multi-language implementation of the quasi maximum likelihood estimator proposed by Aquaro et al. (2021) using MATLAB, Python, R, and Stata.