Show HN: AutoThink – Boosts local LLM performance with adaptive reasoning Y Combinator 2025-05-28 02:39 Source Original site Comments