Qwen 3.8 27B is excellent, but it defaults to overthinking things
kept by eddie
The Qwen 3.8 27B language model delivers strong performance yet frequently produces excessively verbose, over‑analyzed outputs that hinder efficiency in practice.