Solar tower technology offers a promising solution for harnessing solar radiation for power generation. Pressurized gas flows within solar receivers of solar towers are inherently highly turbulent and exhibit significant temperature variations, leading to complex velocitytemperature coupling. These flows are governed by complex low-Mach number NavierStokes equations. The analytical solutions of these flows are intractable, necessitating the use of numerical simulations. Direct Numerical Simulation (DNS) offers the most accurate solutions but comes at a significant computational cost. Large Eddy Simulation (LES) provides a balance between cost and accuracy, resolving large-scale turbulent eddies while modelling the effects of smaller scales. This study investigates the effectiveness of different subgrid-scale (SGS) modelling approaches in capturing the interaction between resolved and unresolved turbulent scales within a turbulent, anisothermal channel flow. A mean friction Reynolds number of approximately 800 was considered. Two nonlinear terms corresponding to velocity–velocity and mass–velocity terms are modelled. Functional, structural, and mixed SGS models were investigated. A parametric study was conducted on one- and two-layer mixed models. The mesh resolution on the model accuracy was investigated. The performance of these models was evaluated by comparing the relative error of the T-LES on mean quantities and correlations and the visualization of the normalized profiles as functions of the wall-normal distance. The results highlight the good agreement of two-layer mixed models consisting of the combination of anisotropic minimum dissipation (AMD) and gradient model (GRD) models with the DNS for the three tested meshes.