HFSS works with THz-frequencies. As PlanarMetamaterials already said, it solves Maxwell's equations, which are valid here.
However, you have to consider (at least) two critical aspects:
The materials in HFSS are usually defined for 10GHz. If you do simulations with gold, silicon, etc. you have to add new materials to the materials database. From my experience, retrieving material properties is the most challenging part in setting up a THz-simulation.
The other point which may lead to simulation problems is the "model unit". The basis unit, used by HFSS, can be chosen from Modeler->Units. Try setting it to "um" or "nm".
We successfully simulated FSS around 300GHz, 1THz and 222THz and propagation in dielectrics up to 2.5THz. The simulation agreed with experiments quite nice.