FACTS benchmark 68.8 score Gemini 3 Pro best factuality model

https://www.foxtrot-bookmarks.win/replacing-hope-with-structure-in-ai-decisions-systematic-ai-validation-for-reliable-enterprise-outcomes

Understanding the Google DeepMind FACTS Test and Its Role in Measuring AI Factuality What the FACTS benchmark evaluates in AI language models As of April 2025, the Google DeepMind FACTS test remains one of the more robust tools to