Claude Opus 4.5 shows 30% on HalluHard - is that low enough?

https://nova-wiki.win/index.php/What_Medical_Review_Boards_Teach_Us_About_Making_AI_Stop_%27Getting_It%27_Wrong

If you have spent any time in the trenches of production AI, you have likely seen the breathless tweets: "Model X beats Model Y on the latest leaderboard." This week, the industry is buzzing about Claude Opus 4