I assume that you are referring to the Normal 2 Mixture and Normal 3 Mixture models available in the Distribution platform. These models are used when the data appear to be a sample from not one but two or three normally distributed populations. The location and scale parameters are estimated for the two or three separate distributions. So the main difference is the number of populations that are modeled. The basis for deciding if and when either one of these models should be used should be clear now.