“ContextFidelity-Bench: A Three-Paradigm Framework for Evaluating Multi-Turn Hallucination Behavior in Language Models”. BenchCouncil Transactions on Benchmarks, Standards and Evaluations 6 (September 30, 2026). Accessed October 8, 2026. https://journal.benchcouncil.org/TBench/article/view/127.