A rating that cannot be checked is a logo. This is the register of every rating in the current edition, dated, searchable, and covering models whether or not their developer engaged. The verification ID on any plaque should bring you to a row on this page.
The record a mark resolves to. The plaque on a slide is a pointer; this row is the claim.
Type a model name, a developer, or the verification ID printed on the mark. Every row states the rating class: SUBSTRATE for a bare model evaluated from public endpoints, CONFIG for a named deployed configuration.
Verification IDs on this demo are generated deterministically from the model name and edition for illustration. In production each ID is issued once, bound to a signed record, and never reused.
A vendor says they’re rated A. The register tells you which model version, in which configuration class, on what date, and whether that rating is still current or was superseded three releases ago.
A rating you can point at is worth more than one you can only assert. The register is where a good result becomes usable in a procurement conversation, and where a stale one gets replaced rather than quietly kept.
Non-participating models are listed too, evaluated from public endpoints. A register that only contained paying customers would be a directory of customers, and worth nothing to a buyer.
The rule that makes this work. Every licensed instance of the mark must link back to its row here: the same requirement UL, the Joint Commission and URAC all impose. The plaque is a pointer; the record is the claim. That is also what makes a counterfeit detectable: a mark with no resolving record is simply false.
Nothing is deleted. A superseded rating stays visible with its successor named, because a buyer looking at a two-year-old slide deck needs to find out that the number on it has moved.
The evaluated version matches what the developer ships, and the benchmark edition is the latest. This is the only state in which the mark may be displayed.
Issued when the model version changes or a new benchmark edition publishes. The record remains, with the successor linked. The mark must come down within five business days.
For misuse of the mark, misrepresentation of scope, or evidence that the evaluated version was not the version shipped. The record stays, and states why.
Name the model, the configuration or the hospital build. You get the price, the calendar and the scenario battery in writing before anything runs. Fees are published in advance and never contingent on a result.