@inproceedings{song-etal-2026-thinkbrake,
    title = "{T}hink{B}rake: Efficient Reasoning via Log-Probability Margin Guided Decoding",
    author = "Song, Sangjun  and
      Oh, Minjae  and
      Lee, Seungkyu  and
      Jo, Sungmin  and
      Jo, Yohan",
    editor = "Liakata, Maria  and
      Moreira, Viviane P.  and
      Zhang, Jiajun  and
      Jurgens, David",
    booktitle = "Findings of the {A}ssociation for {C}omputational {L}inguistics: {ACL} 2026",
    month = jul,
    year = "2026",
    address = "San Diego, California, United States",
    publisher = "Association for Computational Linguistics",
    url = "https://preview.aclanthology.org/ingest-acl/2026.findings-acl.1095/",
    pages = "21765--21790",
    ISBN = "979-8-89176-395-1",
    abstract = "Large Reasoning Models (LRMs) allocate substantial inference-time compute to Chain-of-Thought (CoT) reasoning, improving performance on mathematics, scientific QA, and tool usage. However, this introduces overthinking: LRMs often reach a correct intermediate solution, continue reasoning, and overwrite it with an incorrect answer. We first demonstrate that oracle stopping{---}where we inject lt;/think gt; at every sentence boundary and select the best stopping point in hindsight{---}improves average accuracy by 8{\%} while reducing thinking tokens by 72{\%}, exposing substantial overthinking. Motivated by this finding, we propose ThinkBrake, which monitors the log-probability margin between the top continuation token and lt;/think gt; at sentence boundaries, stopping reasoning when this margin narrows. ThinkBrake requires no training and achieves favorable accuracy{--}efficiency trade-offs across math, scientific QA, and tool usage benchmarks, reducing thinking token usage by up to 30{\%}. Furthermore, we provide theoretical analysis showing that ThinkBrake is equivalent to test-time realignment with a reward bonus for the lt;/think gt; token."
}