The model itself doesn't censor the question.
But if you ask the model via Moonshot as provider, it is their company filter that is blocking it, not the model.
Thus Venice's version of Kimi K3 is superior to Moonshot's own version of Kimi K3 because a) we don't add a layer of censorship on top and b) we implement some light system prompting to encourage models to be more open-minded, shall we say (and we allow users to even turn off our system prompt or change it to whatever they want).