Grain boundaries (GBs) can undergo phase-like transitions that can affect or even control fabrication processing, microstructural development, and a broad range of materials properties. In the last a few years, we have developed several phenomenological and statistical interfacial thermodynamic models to construct GB “phase” (complexion) diagrams. This presentation will be focused on our most recent effort on using atomistic simulations to construct GB diagrams with more details. Specifically, we utilize a hybrid molecular dynamics (MD) and Monte Carlo (MC) method, in conjunction a modified genetic algorithm (GA), to predict equilibrium GB structures and transitions in a semi-grand canonical ensemble. The results from atomistic simulations of Ni-doped Mo and other systems are used to construct various GB diagrams to represent the GB excess, structural disorder, and GB diffusivity as functions of both the bulk composition and equilibrium temperature.