SummEval

A benchmark of 23,325 judgments across the RAGTruth and SummEval human-labeled datasets found that TypeSafe's Jev matched Claude Opus 5 at 87% accuracy on held-out hallucination detection at roughly 1…

A new arXiv paper (2609.10996v1) reports that verbalized confidence has become the more robust soft-scoring mechanism for LLM-as-a-Judge on post-2025 proprietary models, reversing the standard advice …

Researchers propose BINEVAL, a framework that decomposes LLM evaluation into atomic binary questions for interpretable, multi-dimensional scoring. The method matches or outperforms strong baselines on…