Our March 2026 update tracks how leading LLMs handle factual accuracy. We test...

https://dibz.me/blog/perplexity-sonar-pro-citations-is-37-hallucination-the-best-we-have-1070

Our March 2026 update tracks how leading LLMs handle factual accuracy. We test models against the FACTS benchmark to measure error frequency in real-world scenarios. Our data shows current top-tier systems now achieve a hallucination rate as low as 0