PRBench: Large-Scale Expert Rubrics for Evaluating High-Stakes Professional Reasoning

Large language models (LLMs) are increasingly utilized in finance and economics, where their ability to understand chronology is critical. A study tested this capability through various chronological ordering tasks, revealing that while models like GPT-4.1 and GPT-5 can maintain local order, they struggle with creating a consistent global timeline. The findings indicate a significant drop in exact match rates as task complexity increases, particularly in conditional sorting tasks, highlighting inherent limitations in LLMs' chronological reasoning.

Anomaly detection is essential in fields like cybersecurity and finance, particularly with large-scale tabular data. Traditional unsupervised methods struggle due to their reliance on a single global distribution, which does not account for the diverse contexts present in real-world data. This paper introduces a contextual learning framework that models normal behavior variations across different contexts, focusing on conditional data distributions instead of a global joint distribution, enhancing anomaly detection effectiveness.

Large language models (LLMs) are known for their impressive text generation capabilities; however, they frequently produce factually incorrect content, a phenomenon referred to as hallucination. This issue is particularly concerning in critical fields such as healthcare and finance. Traditional methods for detecting these inaccuracies often require multiple API calls, leading to increased latency and costs. The introduction of CONFACTCHECK offers a new approach that checks for consistency in responses to factual queries, enhancing the reliability of LLM outputs without needing external knowled…