We track how AI models perform under pressure. Our March 2026 update evaluates...

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

We track how AI models perform under pressure. Our March 2026 update evaluates the latest LLMs against the rigorous FACTS benchmark to measure real-world reliability. You will find that top-tier models now achieve a hallucination rate of just 0