If you are using an API from a language that supports parallel threading, then yes it can be done, but you need to handle the parallelism. Set the threads parameter in each of your CPLEX models to 1 and spawn each model in a new thread using whatever method your chosen language provides. In Java, I would wrap the code that solves the model in a class that implements the Runnable interface, then run it in a new Thread object.
#CPLEXOptimizers#DecisionOptimization