Grok-3 2.1% Old vs 5.8% New Benchmark xAI Accuracy Decline

https://www.bookmarking-base.win/when-ai-sounds-certain-but-is-wrong-why-zero-hallucination-is-a-myth-and-what-you-can-do-about-it

xAI Grok Performance and the Impact of Benchmark Versions on Hallucination Rates Why Hallucination Rates Vary Across Model Versions As of March 2026, scrutiny around AI hallucination rates has intensified with the release of the