That is more or less what BAML does | Hacker News

Hacker Newsnew | past | comments | ask | show | jobs | submit

		kemiller 18 days ago \| parent \| context \| favorite \| on: Structured outputs create false confidence That is more or less what BAML does

refulgentis 18 days ago [–]

I understand this but A) then they should have done it here B) the idea that you can't get CoT x JSON without sacrificing JSON formatting is flat out wrong with ~any 2025 model. (i.e. reasoning models and their APIs specifically enable this)

Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact