Maybe, if I took the time to look at the web page listed above, I would have seen that it is the latest version, AND that is what is offered, AND the release version is given.
By-the-way, I got the CHARMM <---> Gaussian interface running for the latest release of CHARMM. I ran it on a transition metal complex, QM--> Fe(BPY)_3 2+ (Fe, tzvp, ligands: 6-31g*) + MM--> 557 H2O molecules + ions for neutralization with RHDO) on one node with 24 cores. I have a job pending for 2 nodes/48 cores. The G09 input has the line %NProcShared=24 (i.e. one node). I am hoping the two nodes will work on the CHARMM (MM) component.
I can supply the fixes, if you want.