Data as of Jul 25, 2026 · Based on 2,718,867 AI responses across 9,511 prompts · See how Parse measures this
Factcheck-GPTgithub.com · Low evidenceFactcheck GPT is a framework for document-level fact-checking of generative large language model outputs, covering decomposition, decontextualization, check-worthiness identification, evidence retrieval, stance detection, and error correction. It includes a human-annotated dataset of 94 ChatGPT responses and a larger FactBench dataset of 4,835 examples for evaluating automatic fact checkers.
Parse Score
Sources
aclanthology.org shapes more of what AI says about Factcheck-GPT than any other source, at 20% of its citations.
arxiv.org · evidentlyai.com · pmc.ncbi.nlm.nih.gov · prompts.ai