Résumé : In this paper, a gradient-based refinement of the rotation rate-based Smagorinsky (RoSM) subgrid scale (SGS) model is presented. The refined model satisfies the Galilean invariance condition generally, without any assumptions. The suggested model retains the advantages offered by the original RoSM, thus being simple and efficient. It provides a Smagorinsky model constant that is always positive, with low fluctuations in space and time, without the need for any numerical stability control algorithms. The validity of the proposed SGS model is shown through three test cases, namely, a turbulent channel flow, subcritical flow past a stationary cylinder, and a spatially developing free round jet. The refined RoSM provides comparable results with the dynamic Smagorinsky, while matching well to reference data. The refined RoSM is shown to be computationally efficient, being 20% faster than the dynamic Smagorinsky model.