© 2017 Elsevier B.V. Using a basin-hopping global optimization method, an exhaustive global search of the potential energy surface (PES) of trimetallic Ag-Au-Pt clusters is performed at the empirical potential (Gupta) level. The clusters contain a total of N = 13, 19, 33, 38 atoms while the Ag, Au, Pt content range from 0% to 100%. A broad region of the PES of these systems is sampled, finding a menagerie of geometries, which are catalogued into structural families according to geometrical considerations. The structural families so defined contain a fixed number of Pt atoms while the loading of Ag and Au atoms varies, and indicate that selective Pt doping can stabilize Ag-Au subnanometer clusters. Structural stability trends are identified thus obtaining the optimal composition for each cluster size. Segregation is analyzed through a chemical order parameter, and found to exhibit a size-dependent behavior, with a transition from mixed to segregated structures clearly evidenced at 38-atom cluster size. Optimal composition structures of the Gupta global minima and a few isomers are subjected to DFT reoptimization, which is found to predict qualitatively similar atomic arrangements with a certain degree of deformation, especially for 38-atom clusters, and to introduce changes in the energy ordering among structures with respect to the Gupta potential predictions, although the stabilizing effect of Pt and a cross-over character of 38-atom size seem confirmed.