ParentFull threadentropicdrifter·So you mean something like, "what if the baseline, off-the-cuff response for the next-gen models was tuned based on the results of the reasoning model excluding the reasoning itself?"View on HN