AA-Omniscience is Artificial Analysis' knowledge-reliability and hallucination benchmark with 6,000 questions across 42 topics and 6 major domains. It reports an AA-Omniscience Index from -100 to 100 that rewards correct answers, penalizes incorrect answers and treats abstention neutrally, alongside Accuracy and Hallucination Rate.
Browse the latest scores, model modes, release dates, and parameter sizes for AA-Omniscience.
Data sourced primarily from official releases (GitHub, Hugging Face, papers), then benchmark leaderboards, then third-party evaluators. Learn about our data methodology
No benchmark data available yet
