[skynotblue]

Would love to see how smaller distilled models perform with this agent enabled.

Frontier models already had all the Q/A in their training dataset and there's not much activity there anymore so very little use for searching new Q/A.