Solver not giving optimal solution

Any notable difference in the fixed final solve and the MIP solution usually indicates some numerical issue in the model (e.g. poor scaling). See the following thread:
MIP solution and final solve are different