Model Card
The documentation shipped with a model.
A standard description of a model: intended use, training data, evaluation results, limitations, and known risks. A disclosure artifact for users and auditors.
Read them for the limitations section. It is usually the most honest and most useful part, and it rarely makes the marketing page.
Its practical function is to shift information from the vendor to the consumer before a decision, which is why intended use and limitations matter more than headline scores. Read it for the caveats: what the model was not evaluated on is usually the most useful section.
Reading the limitations first
Selecting a model for a regulated use case.
The intended-use section tells you whether your use is in scope, and the limitations tell you which evaluations were not run. A model that documents honestly is easier to defend in a review than one that only publishes benchmark wins.
- standardAI Risk Management Framework: Generative AI Profile (NIST AI 600-1) ↗
The official record for the cross-sector GenAI companion to the AI RMF. Use it as the governance vocabulary; the PDF itself was not machine-readable in this research pass, so do not quote definitions from here.
- reportAnnouncing our updated Responsible Scaling Policy ↗
Capability thresholds with graduated safeguards and documentation duties, written in the safety-case style: here is the evidence that this risk is adequately handled.
Content Policy
The written rules about what the system may produce.
Evaluation · Evals
How you know the system is actually working.
Data Governance
Rules about what data may be used, by whom, and retained how long.
Red Teaming
Adversarially trying to make the system do the wrong thing.