The model itself is hardware-agnostic, so there's nothing preventing someone from building a frontend for their platform of choice. Granted, powerful hardware

The model itself is hardware-agnostic, so there's nothing preventing someone from building a frontend for their platform of choice. Granted, powerful hardware is still required to run inference at acceptable speeds (or at all - I don't know the memory requirements).