arXiv AI August 12, 2026 By Dahai Yu, Lin Jiang, Rongchao Xu, Guang Wang

TrAC: Trace-Conditioned Answer Consistency for Efficient Uncertainty Quantification in LLMs

arXiv:2608. 00422v2 Announce Type: replace Abstract: Large language models (LLMs) can generate fluent reasoning traces that nevertheless lead to incorrect answers, making response-level uncertainty estimation important for abstention, human review, and adaptive compute allocation.

Summary generated by The Flow from the publisher's feed. The full article lives at arXiv AI.

arXiv AI

Jun 3

Quantifying Faithful Confidence Expression in Large Reasoning Models

arXiv:2606. 03969v1 Announce Type: cross Abstract: Reliable uncertainty communication is critical to the trustworthiness of LLMs, yet faithful calibration (FC)--the alignment between models' intrinsic and (linguistically) expressed confidence--is a persistent failure mode.

By Areeb Gani, Asal Meskin, Gabrielle Kaili-May Liu, Arman Cohan

llms safety

arXiv AI

Jun 9

Diagnosing Multi-step Reasoning Failures in Black-box LLMs via Stepwise Confidence Attribution

arXiv:2605. 19228v2 Announce Type: replace-cross Abstract: Large Language Models have achieved strong performance on reasoning tasks with objective answers by generating step-by-step solutions, but diagnosing where a multi-step reasoning trace might fail remains difficult.

By Xiaoou Liu, Tiejin Chen, Dengjia Zhang, Yaqing Wang, Lu Cheng, Hua Wei

llms nlp

arXiv AI

Jul 14

Filtered Reasoning Score: Evaluating Reasoning Quality on a Model's Most-Confident Traces

arXiv:2604. 11996v2 Announce Type: replace-cross Abstract: Should we trust Large Language Models (LLMs) with high accuracy?

By Manas Pathak, Xingyao Chen, Shuozhe Li, Amy Zhang, Liu Leqi

llms benchmarks

arXiv AI

Jun 30

Does Verbose Chain-of-Thought Really Help? In-Distribution Evidence that Content, Not Length, Matters

arXiv:2606. 30128v1 Announce Type: new Abstract: Chain-of-thought (CoT) prompting improves LLM reasoning, but the source is contested: do the intermediate steps help because they carry useful semantic content, or because conditioning on more tokens buys extra computation before the model commits to an answer?

By Wenlong Wang, Fergal Reid

llms benchmarks

arXiv AI

Jul 31

A Reference-Free Score for Detecting Silent Reasoning Failures in Large Language Models

arXiv:2607. 26102v1 Announce Type: cross Abstract: Mathematical chain of thought (CoT) evaluation is commonly reduced to whether the final answer matches a reference.

By Vivek Shukla, Varun Shukla, Atul, Divya Mishra, Mehul Kumar Das

llms

arXiv AI

Jun 8

CORE: Contrastive Reflection Enables Rapid Improvements in Reasoning

arXiv:2605. 28742v2 Announce Type: replace Abstract: Language models can use verifiable rewards to improve at a wide variety of reasoning tasks.

By Linas Nasvytis, Simon Jerome Han, Ben Prystawski, Satchel Grant, Noah D. Goodman, Judith E. Fan

llms rag

Related stories

Quantifying Faithful Confidence Expression in Large Reasoning Models

Diagnosing Multi-step Reasoning Failures in Black-box LLMs via Stepwise Confidence Attribution

Filtered Reasoning Score: Evaluating Reasoning Quality on a Model's Most-Confident Traces

Does Verbose Chain-of-Thought Really Help? In-Distribution Evidence that Content, Not Length, Matters

A Reference-Free Score for Detecting Silent Reasoning Failures in Large Language Models

CORE: Contrastive Reflection Enables Rapid Improvements in Reasoning