I am using gromacs and the protein I use is a homodimer and both dimers bind from the same region to receptor, to calculate the free energy of binding do i need to use the dimer as input or the monomer is enough?
You need to model whatever is biologically relevant. Some proteins are functional dimers, some are only crystallographic dimers. You need to do the simulation based on what exists in a real biological system.
As Dr. Lemkul said, it depends totally on the protein complex you are working with. For example, insulin remains as a hexamer in a biological assembly. However, insulin binds to its receptor as a monomer. Contrary to this, some growth hormones binds to two receptors together as a dimer.