I think it's better in all cases simply to say that strong Sapir-Whorf is wrong, but that weaker versions have some utility as a means to understand how humans think and speak.