hard to say that as some have mentioned everybody has their own implementation of how to address the 32 bits dynamic range in terms of multiple processors in parallel etc. It would be fairly easy to gear it towards your typical Mic sensitivities and cover the whole range up to max SPL but if you were putting anything with a line signal in I would expect to pad it to mic level or consumer unbalanced line level
again it’s a black box at this point hard to say