AI & provider disclosure
How multiple model services fit behind a Noemad profile.
This page documents preview behavior and launch requirements. Final operator details, legal terms, contact channels, and production commitments have not been established.
One product, multiple adapters
Noemad’s public interface owns the run contract, wallet identity, review flow, and receipt. Configured adapters can connect to different model services without presenting a single upstream brand as the product.
Catalog versus availability
The model library contains archived capability and rate references. Model names and developer names identify the record; inclusion is not an endorsement, affiliation, live connection, or guarantee of availability.
Data sent for a run
A production profile must disclose which input and metadata are sent to the configured provider, why they are needed, and which retention and training terms apply. Do not assume every provider follows the same policy.
Provider changes
Changing a model, adapter, limit, or price creates new future terms. Already accepted runs must retain their immutable configuration and payment snapshot. Unknown outcomes require reconciliation before fallback or duplicate dispatch.
Result limitations
AI output is probabilistic and may contain errors. A result is accepted only after complete response parsing and profile validation; HTTP success alone does not establish a valid answer.