Don't trust generic "99% accurate" claims. Hallucination metrics vary wildly...

https://dibz.me/blog/facts-benchmark-scores-why-is-nobody-above-70-overall-1154

Don't trust generic "99% accurate" claims. Hallucination metrics vary wildly depending on the test. If you use Vectara's HHEM to measure grounding, you see one reality; apply AA-Omniscience for logic, and the picture shifts entirely