The paper investigates token‑level certainty as a proxy for correctness in large language models. It finds that certainty better predicts whether a model will answer a question correctly than it does whether a specific response is correct, and that certainty varies by token type and position. The authors show that using certainty early in generation to allocate responses and later to weight votes improves accuracy while dramatically cutting token cost.
By Yunfan Zhou, Ye Zhu, Zhihai Wang, Jianguo Yao, Haibing Guan, Xijun Li
The paper introduces the concept of LLM‑specific utility, defining it as the performance gain a target large language model (LLM) achieves when provided with a passage compared to answering without evidence. A benchmark of utilitarian passages is built for four LLMs (Qwen3‑8B/14B/32B and Llama 3.1‑8B) across three QA datasets, revealing that each model benefits most from its own tailored evidence and that evidence optimized for other models is consistently suboptimal. The authors also create SpecUBench, a benchmark for LLM‑specific utility judgment, and show that current utility‑aware retrieval methods largely capture model‑agnostic usefulness, struggling to estimate LLM‑specific utility.
"whyItMatters":"The study demonstrates that retrieval‑augmented generation must consider model‑specific evidence selection to truly improve LLM performance, highlighting a gap in existing utility‑aware methods."
By Hengran Zhang, Keping Bi, Jiafeng Guo, Jiaming Zhang, Shuaiqiang Wang, Dawei Yin, Xueqi Cheng
arXiv:2607. 25600v1 Announce Type: cross Abstract: Retrieval-augmented generation improves knowledge-intensive question answering, but indiscriminate retrieval can introduce irrelevant evidence and unnecessary computation.
By Chandan Kumar Sah, Xiaoli Lian, Li Zhang
The paper introduces “Rel-Bloom,” a multilingual diagnostic benchmark of 450 multiple-choice questions that tests language models’ ability to reason over relational facts supplied directly in prompts. It evaluates 43 models across six languages and two multilingual settings, revealing that accuracy drops sharply from Apply to Analyze levels, that Evaluate items remain a bottleneck even for high-performing models, and that difficulty rankings are consistent across languages. The study also finds that counterfactual contexts are generally harder than factual ones, and that removing entity identities can sometimes improve higher-level performance.
By Zohreh Abbasi Liasi, Batoul Hammoud, Anas Madkoor, Ehsaneddin Asgari
PROOF is a benchmark that profiles the reliability of object-level facts in instruction-tuned language models by converting a frozen Wikidata snapshot into 18,486 English multiple-choice questions grounded in 11,779 semantic facts across 101 classes, 392 properties, and 14 domains. Each question includes an explicit "I don't know" option, a "No correct option" control, and nine controlled formulations, with 1,849 questions designed as no-correct-option traps. The study evaluates 18 open-weight model deployments on 166,374 prompts, revealing wide variability in factual accuracy, sensitivity to wording changes, and the impact of decoder perturbations.
By Andrei Chetvergov, Mikhail Solovev, Timofei Sivoraksha, Stepan Ukolov, Valeriia Kuschenko, Alexander Evseev, Sergey Bolovtsov
Retrieval-augmented generation improves knowledge-intensive question answering, but indiscriminate retrieval can introduce irrelevant evidence and unnecessary computation. We investigate whether verbalized confidence from black-box language models can serve as an actionable signal for retrieval routing.
The paper introduces a penalty‑aware evaluation framework for Retrieval‑Augmented Generation (RAG) systems that uses asymmetric scoring, knowledge‑gap canaries, and a failure‑attribution pipeline. Applying this framework to three commercial RAG products and a baseline on SimpleQA‑Verified, the authors find that while overall accuracy is similar across systems, canary violation rates vary dramatically, showing that systems differ more in when they answer than in what they answer. The study demonstrates that penalty‑aware scoring can reorder system rankings and is robust across different penalty settings.
By Alden Do Rosario, Hussein Younes, Felipe Pires
ConfRAG introduces a confidence-guided approach to reduce hallucinations in large language models and selectively trigger Retrieval-Augmented Generation (RAG) only when the model is uncertain. The ConfQA fine‑tuning strategy trains the model to answer correctly or respond with "I am unsure," achieving a drop in hallucination rates from 20‑40% to below 5% across factuality benchmarks. Building on ConfQA, ConfRAG limits external retrievals by more than 30% while maintaining over 95% accuracy in ideal scenarios.
By Yin Huang, Yifan Ethan Xu, Kai Sun, Vera Yan, Alicia Sun, Haidar Khan, Jimmy Nguyen, Jingxiang Chen, Mohammad Kachuee, Zhaojiang Lin, Yue Liu, Aaron Colak, Anuj Kumar, Wen-tau Yih, Xin Luna Dong
arXiv:2509.24988v2 Announce Type: replace-cross
Abstract: Generating accurate and calibrated confidence estimates is critical for deploying LLMs in high-stakes or user-facing applications, and remain...
By Hanqi Xiao, Vaidehi Patil, Hyunji Lee, Elias Stengel-Eskin, Mohit Bansal
arXiv:2607. 05861v1 Announce Type: cross Abstract: Large reasoning models (LRMs) improve language model capabilities by generating explicit thinking traces before final answers.
By Kaishen Wang, Tong Zheng, Xuehao Cui, Ruibo Chen, Tianyi Xiong, Heng Huang
The study investigates sycophancy in Chinese large language models (LLMs) by analyzing 364,941 responses from DeepSeek, Qwen, and Doubao to 12,165 yes/no factual questions derived from real-world search queries. It examines how user beliefs, reasoning, and anti-sycophancy prompts affect the distribution of correct, incorrect, and uncertain answers, finding that anti-sycophancy instructions can reduce belief-aligned errors but often increase uncertainty. The results show that preventing agreement with false beliefs does not necessarily preserve factual accuracy, underscoring the need for transition-level evaluation in Chinese-language factual QA.
By Geng Liu, Feng Li, Mengxiao Zhu, Francesco Pierri
The paper investigates why large language models often answer factual questions incorrectly in closed‑book settings. It shows that many errors are due to first‑token recall failures: the model commits to an incorrect first token but can complete the correct answer once the right token is provided. Using a First‑Token Cueing probe, the authors demonstrate that correcting just the first token recovers a significant portion of errors across 16 open models and improves accuracy on multilingual benchmarks.
By Jaewon Sok, Jewon Yeom, Seonghyeon Park, Jeongjae Park, Hwiyeong Lee, Taesup Kim