Before answering to your question, I must say that computing accurate absolute, or even relative, binding energies is an extremely challenging task. The GlideScore (docking score) is an estimate of the binding affinity that can distinguishing actives from inactives. However, from the binding energy calculations viewepoint, Prime MM-GBSA and MacroModel Embrace are two programs that allow receptor flexiblity, and you can have promising results in particular with Prime MM-GBSA for rank-ordering congeneric series of ligands.
Alternatively, one can compute the Total Free Energy by FEP panel (molecular dynamics simulation with Desmond) by selecting In complex in the Plan Calculation tab. With this, you can also set up jobs to mutate ligand structures and protein structures too.