Turbulent combustion modelling at the level of the one-point joint velocity-scalar PDF requires models for the evolution of the joint statistics in velocity space. In a RANS framework when using a particle stochastic approach, the Generalised Langevin Model (GLM) can be used. In the 1990's, it was shown that such models can be chosen to be in correspondence with a given Reynolds-stress turbulence model [1,2]. Moreover, a mixing model is required in order to close the evolution of the joint PDF in scalar space. The combination of a mixing model with a GLM implies a given model for the velocity-scalar correlation evolution (i.e. a given scalar-flux model). Recently, a GLM formulation was proposed in order to be in correspondence with a given scalar-flux model depending on the chosen mixing model [3,4]. In particular, a variable Cφ* was introduced in order to represent the contribution of the mixing model to the pressure-scrambling term in the implied scalar-flux modelled transport equation. The value of Cφ* is zero when using mixing models designed to be local in velocity space. When using the simple but widely used LMSE mixing model [5], it is easy to show that Cφ*=Cφ/2 (with Cφ the mixing model constant). However, its value is more difficult to determine a priori in more complex mixing models. We evaluate here the value of Cφ* for different mixing models in turbulent flame calculations and include a proper value in the GLM in order to yield the desired scalar-flux model. [1] S.B. Pope. Phys. Fluids, 6:973-985, 1994. [2] H.A. Wouters, T.W.J. Peeters and D. Roekaerts. Phys. Fluids, 8:1702-1704, 1996. [3] B. Naud, B. Merci and D. Roekaerts. Flow Turbul. Combust., 85:363-382, 2010. [4] B. Naud and D. Roekaerts. Phys. Fluids, 33:035101, 2021. [5] C. Dopazo and E.E. O’Brien. Acta Astronautica, 1:1239-1266, 1974.