Text this: Performance of large language models (GPT-5.2, Gemini 3 Pro, Claude Sonnet 4.6 and Grok 4.1) on the Fellowship of The Royal College of Surgeons Urology Part A examination.