Content checks
Some models on kenari run an automatic content check on each request before it reaches the model. This page covers which models, what is not allowed on them, and what happens when a request is blocked.
Which models
Section titled “Which models”The Muse models and the GPT models. Other models are not checked this way, but the Terms of Service apply on every model.
What is not allowed
Section titled “What is not allowed”On the checked models, a request that breaks our Terms or the model provider’s usage policy is blocked. That includes sexually explicit content, sexual content involving minors, and help with attacking systems or writing malware.
Sexual content involving minors is never allowed on any model, in fiction and role-play too.
We do not publish the exact criteria, so the checks cannot be worked around.
When a request is blocked
Section titled “When a request is blocked”A blocked request fails with HTTP 400 and the code request_blocked. A refusal from the model provider itself counts the same way. See Errors.
When an account’s requests are blocked again and again in a short time, the account is paused on the checked models for a while. During the pause, those models answer with HTTP 400 and the message this model is temporarily unavailable for this account after repeated blocked requests. The pause ends by itself, and every other model keeps working.
If a block looks wrong
Section titled “If a block looks wrong”Open the Help menu in the dashboard and create a ticket with the time of the request. We look at it and reply in the ticket.