Different upstream providers treat data differently, and the same provider can behave differently depending on the model or account type. To make this visible, the administrator answers a fixed set of YES/NO questions for every model offered through SaaufAI.
How to read the answers
- Yes: the statement applies to this model.
- No: the statement does not apply.
- Not specified: the administrator has not answered this question yet. Treat it as unknown and assume the less private option.
Answers may come with a short note, for example a retention period or a link to the provider's terms.
Automatic answers
Two questions are answered automatically from the gateway's live configuration, so they can never be out of date: whether SaaufAI enforces its own safeguards on the model (and which limits), and whether requests may be answered by a fallback model (and which ones). A fallback model may run at a different provider with different data terms, so check its own disclosures too. Responses served by a fallback still carry the requested model id.
Where to see them
The full matrix is on the Models page, and every model has its own page with the answers and notes. Answers reflect the administrator's best understanding of the upstream provider's terms at the time they were written; providers can change their terms, and the provider's own documentation always takes precedence.
What SaaufAI itself does
Regardless of the model, SaaufAI itself does not store prompt or response content. See the Privacy Policy.