In recent years, a new class of super saturated binary and ternary alloys have demonstrated the ability for the self-healing of creep-induced voids formed at the grain boundaries. However, a clear understanding of the parameters affecting the self-healing mechanism is still not yet complete. One of the main challenges is understanding the effect of microstructure and micromechanical stresses on the redistribution of the healing-solute and vacancies. To this end, we address this issue using a CALPHAD-informed diffusion model coupled with crystal plasticity. In principle, the approach is general and can be used for any binary Fe–X alloy, but in this work Fe–Au binary system is used since it experimentally showed the best healing efficiency. First, we present a multicomponent diffusion model considering cross and stress-driven diffusion. The effect of stress was also considered on the equilibrium vacancy concentration. To investigate the effect of the micromechanical stresses, a representative volume element (RVE) was obtained using the phase-field method. The results showed that the maximum vacancy concentration is at the grain boundaries (GBs) with the highest hydrostatic tensile stresses. These were also the regions of the highest Au enrichment. A crucial factor to achieve this is the high diffusivity of Au compared to the Fe matrix. Increasing the stresses, lead to an increase both in vacancy and Au concentration. The accompanying increased stress triaxiality is suggested to be the reason for the reduced self-healing efficiency observed in previous experimental studies.