As the first step, try to compare model statistics in engine logs. If the engine logs are different, the two models might be different. Even if they have the same statistics, if the ordering of constraints or variables are different, you might still see different solutions.
However, as long as the two CPLEX models are indeed mathematically equivalent, you should expect the same or nearly equal objective values.
If you need further discussion, please open a case if you are entitled. Otherwise, please discuss your question in Data Science Community https://community.ibm.com/community/user/datascience/communities/community-home/digestviewer?communitykey=ab7de0fd-6f43-47a9-8261-33578a231bb7&tab=digestviewer
#DecisionOptimization#Support#SupportMigration