LLMs Score High on FRCR Physics, Struggle on Anatomy in Comparative Study

Claude led physics (195/200) but Gemini led anatomy (160/200) on an FRCR Part 1-style exam. LLMs performed well on text but showed variable accuracy on image-based anatomy, highlighting educational potential and limitations.

• Three frontier LLMs (GPT-5.4 Thinking, Gemini 3.1 Pro, Claude Opus 4.6) were tested on a custom 200-statement physics exam and 100 image-based anatomy questions.
• Physics scores: Claude 195/200, GPT 189/200, Gemini 187/200. Claude significantly outperformed Gemini (adjusted P=0.023).
• Anatomy scores: Gemini 160/200, GPT 137/200, Claude 83/200; all pairwise comparisons significant (adjusted P≤0.038).

Automated summary

RadPigeon summaries are original and for information only. They are not clinical advice.