It may be a bad bench, but the three going Google models are measured as having the lowest hallucination rates of any llms in history]
https://artificialanalysis.ai/evaluations/omniscience?models...