In 2026, "hallucination rate" is a vanity metric if you don’t define the test....

https://lukasnpyy234.cavandoragh.org/claude-vs-gpt-which-is-better-at-admitting-i-don-t-know

In 2026, "hallucination rate" is a vanity metric if you don’t define the test. Comparing benchmarks like Vectara’s HHEM against AA-Omniscience is apples to oranges; one measures factual grounding while the other tests logic