Reasoning Jury: Multi-Model Consensus for Evaluating Reasoning Traces

• URL: http://arxiv.org/abs/2608.12585v1
• Date: Wed, 12 Aug 2026 21:00:22 GMT
• Title: Reasoning Jury: Multi-Model Consensus for Evaluating Reasoning Traces
• Abstract summary: Reasoning Jury is a system that replaces the single judge with a jury of LLMs.<n>In Reasoning Jury, defects of a reasoning trace and their severity are surfaced through a deliberation.<n>We show that Reasoning Jury with a jury of open-weight models is able to significantly outperform frontier models.
• Score: 8.727904788574602
• License: http://creativecommons.org/licenses/by/4.0/
• Abstract: Improving reasoning LLMs requires the ability to judge the quality of long reasoning traces for effective reasoning data curation, strong training signals during reinforcement learning, and an in-depth understanding of reasoning behaviors during model performance evaluation. Additionally, surfacing reasoning mistakes that the model makes would enable improving the model's performance at runtime through providing feedback. Due to the difficulty of this complex task on long reasoning traces, single-model judges (even frontier models) do not do well at identifying reasoning defects. Additionally, leveraging frontier models during online training of reasoning LLMs is generally prohibited due to guardrails in terms of use. In this work, we introduce Reasoning Jury, a system that replaces the single judge with a jury of LLMs and a moderated consensus mechanism, to improve the fidelity of judgments for identifying reasoning defects. In reasoning jury, defects of a reasoning trace and their severity are surfaced through a deliberation where a moderator conducts a discussion amongst the jury where the jurors critique each other's judgments and get to modify their initial votes. The moderator derives a consensus through deliberation amongst jurors or consolidation of judgements. We show that Reasoning Jury with a jury of open-weight models (e.g., gpt-oss-120b) is able to significantly outperform frontier models (opus-4.6, sonnet-4.6, and gemini-3.1-pro) at correctly identifying reasoning defects. Besides accuracy performance improvements, the aggregated cost of the jury (initial verdicts, deliberations, consolidation, etc.) is a fraction (8 to 15%) of the cost of running frontier models in LLM-as-a-judge setup. We also show how these judgements can be leveraged to understand failure modes of reasoning LLMs on benchmarks, which allows much deeper understanding of a model's performance.

Related papers

• AffectOmni: RL-Verifiable People-Centric Grounded Affective Reasoning for Social and Art-Related Scenes [59.134503227559634]
  Multimodal large language models (MLLMs) achieve strong performance on VQA and scene understanding.<n>Models may predict correct answers while neglecting people-centric cues such as micro expressions and body language.<n>We propose Affect Omni, a framework for verifiable affective reasoning.
  arXiv  Detail & Related papers  (2026-08-24T11:25:04Z)
• How Long Reasoning Chains Influence LLMs' Judgment of Answer Factuality [9.19183567561999]
  Large language models (LLMs) has been widely adopted as a scalable surrogate for human evaluation, yet such judges remain imperfect and susceptible to surface-level biases.<n>With the rise of reasoning-capable models, exposing a generator's reasoning content to the judge provides richer information and is a natural candidate for improving judgment accuracy.<n>We find that weak judges are easily swayed by reasoning presence, while strong judges can partially leverage reasoning as informative evidence.
  arXiv  Detail & Related papers  (2026-04-08T07:21:18Z)
• Examining Reasoning LLMs-as-Judges in Non-Verifiable LLM Post-Training [75.98427023288052]
  Reasoning LLMs-as-Judges provide a promising path for extending the success of reasoning models to non-verifiable domains.<n>This study investigates the actual impact of non-reasoning and reasoning judges in reinforcement-learning-based LLM alignment.<n>We find that reasoning-judge-trained policies achieve such strong performance by learning to generate highly effective adversarial outputs.
  arXiv  Detail & Related papers  (2026-03-12T17:57:06Z)
• Adversarial Yet Cooperative: Multi-Perspective Reasoning in Retrieved-Augmented Language Models [72.4149653187766]
  We propose a Reasoner-Verifier framework named Adrialversa Reasoning RAG (ARR)<n>The Reasoner and Verifier engage in reasoning on retrieved evidence and critiquing each other's logic while being guided by process-aware advantage.<n> Experiments on multiple benchmarks demonstrate the effectiveness of our method.
  arXiv  Detail & Related papers  (2026-01-08T06:57:03Z)
• Who Judges the Judge? LLM Jury-on-Demand: Building Trustworthy LLM Evaluation Systems [2.9141470183751674]
  We propose a dynamic, learning-based framework for scalable and context-aware evaluation.<n>Our method trains a set of reliability predictors to assess when LLM judges will agree with human experts.<n> Experiments on summarization and RAG benchmarks show that our dynamic jury system achieves significantly higher correlation with human judgment than both single-judge and static-jury baselines.
  arXiv  Detail & Related papers  (2025-12-01T15:26:20Z)
• Thinking Longer, Not Always Smarter: Evaluating LLM Capabilities in Hierarchical Legal Reasoning [11.255428720705204]
  We propose a framework that decomposes the process of identifying significant distinctions between cases into three-stage reasoning tasks.<n>Our framework models cases using factual predicates called factors, organizes them into a legal knowledge hierarchy, and defines verifiable rules for identifying distinctions.<n>We find that models consistently expend more computational resources on incorrect responses than correct ones, suggesting that "thinking longer" does not always mean "thinking smarter"
  arXiv  Detail & Related papers  (2025-10-09T18:15:28Z)
• J4R: Learning to Judge with Equivalent Initial State Group Relative Policy Optimization [69.23273504123941]
  We train judges to be robust to positional biases that arise in more complex evaluation settings.<n>We introduce ReasoningJudgeBench, a benchmark that evaluates judges in diverse reasoning settings not covered by prior work.<n>We train Judge for Reasoning (J4R), a 7B judge trained with EIS-GRPO that outperforms GPT-4o and the next best small judge by 6.7% and 9%.
  arXiv  Detail & Related papers  (2025-05-19T16:50:35Z)
• Reasoning Court: Combining Reasoning, Action, and Judgment for Multi-Hop Reasoning [17.829990749622496]
  Reasoning Court (RC) is a novel framework that extends iterative reasoning-and-retrieval methods, such as ReAct, with a dedicated LLM judge.<n>RC consistently outperforms state-of-the-art few-shot prompting methods without task-specific fine-tuning.
  arXiv  Detail & Related papers  (2025-04-14T00:56:08Z)
• JudgeLRM: Large Reasoning Models as a Judge [80.07261839142548]
  We introduce JudgeLRM, a family of judgment-oriented Large Language Models (LLMs)<n>We find a negative correlation between SFT performance gains and the proportion of reasoning-demanding samples, revealing the limits of SFT in such scenarios.<n>We show that JudgeLRM consistently outperform SFT-tuned baselines in the same size, as well as other RL and SFT variants, and even surpass state-of-the-art reasoning models.
  arXiv  Detail & Related papers  (2025-03-31T02:18:51Z)
• Judge as A Judge: Improving the Evaluation of Retrieval-Augmented Generation through the Judge-Consistency of Large Language Models [68.92020689188887]
  Retrieval-Augmented Generation (RAG) has proven its effectiveness in alleviating hallucinations for Large Language Models (LLMs)<n>Existing automated evaluation metrics cannot fairly evaluate the outputs generated by RAG models during training and evaluation.<n>This paper introduces the Judge-Consistency (ConsJudge) method, which aims to enhance LLMs to generate more accurate evaluations for RAG models.
  arXiv  Detail & Related papers  (2025-02-26T04:50:43Z)
• JudgeBench: A Benchmark for Evaluating LLM-based Judges [61.048125269475854]
  JudgeBench is a benchmark for evaluating LLM-based judges on challenging response pairs spanning knowledge, reasoning, math, and coding.<n>Our comprehensive evaluation on a collection of prompted judges, fine-tuned judges, multi-agent judges, and reward models shows that JudgeBench poses a significantly greater challenge than previous benchmarks.
  arXiv  Detail & Related papers  (2024-10-16T17:58:19Z)
• From Calculation to Adjudication: Examining LLM judges on Mathematical Reasoning Tasks [11.01213914485374]
  We study large language models (LLMs) on mathematical reasoning tasks.<n>Our analysis uncovers a strong correlation between judgment performance and the candidate model task performance.<n>As a consequence, we test whether we can predict the behavior of LLM judges using simple features such as part-of-speech tags.
  arXiv  Detail & Related papers  (2024-09-06T10:09:41Z)

This list is automatically generated from the titles and abstracts of the papers in this site.

This site does not guarantee the quality of this site (including all information) and is not responsible for any consequences.