Answer Generation Over Docs

Which models are strong at this?

Per-use-case best-fit scoring is surfaced through the modelside of the catalogue: open any model detail page and the Use case fit grid lists the use cases that model scores Strong / Acceptable / Avoid against. The reverse query — for this use case, rank every model— lands once the Phase D Classifier agent pipeline finishes the per-(model, use_case) scoring across the catalogue.