I would imagine Gruber made certain that he controls the source code and the best way to do that is for him to have written the code himself and then keep it to himself. It is reportedly only 15k lines of source code. I would imagine it is coded in a common language such as C, C++, or Fortran, to aid portability to arbitrary computers. One thing the WH boasted when it no-bid contracted with the Groob was his ability to perform quick turnaround on new input parameters. Therefore it is conceivable that runs of the code could take just a few hours or even just a few minutes on a mid-size Mac laptop.
One more thing— if Gruber claims to have written the modeling software on his own time, then Gruber owns the modeling software. If Gruber wrote it on MIT time, then imaginably the software belongs to MIT, not Gruber. From an outsider perspective, the difference might be relatively minor since both Gruber and MIT are private (in contrast to public) entities, and the public is not entitled to see the software (not to mention audit the software for accuracy).