{"id":"W4417080179","doi":"10.1016/j.xops.2025.101034","title":"Performance of GPT-5 Frontier Models in Ophthalmology Question Answering","year":2025,"lang":"en","type":"article","venue":"Ophthalmology Science","topic":"Artificial Intelligence in Healthcare and Education","field":"Medicine","cited_by":2,"is_retracted":false,"has_abstract":true,"ca_institutions":"University of Toronto; Hôpital Maisonneuve-Rosemont; Université de Montréal; Centre Hospitalier de l’Université de Montréal","funders":"Alcon Research Institute; National Institutes of Health; Apellis Pharmaceuticals; American Academy of Ophthalmology; Cleveland Clinic; Moorfields Eye Charity; Roche","keywords":"Question answering; Frontier; Questions and answers; Component (thermodynamics)","routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"about_ca":false,"invisible_to_affiliation_only":false},"retraction":null,"screen":null,"direct_labels":[],"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.007320276,0.001246852,0.0007196541,0.0009193676,0.0005163815,0.001861098,0.003029341,0.001668727,0.004436072],"category_scores_gemma":[0.02730634,0.0004738414,0.001482326,0.0006674745,0.0008886669,0.003351155,0.002672191,0.002864926,0.001325896],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.002814478,"about_ca_system_score_gemma":0.002532025,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.01425071,"about_ca_topic_score_gemma":0.01601951,"domain_scores_codex":[0.9967354,0.001520316,0.0002427403,0.0007588813,0.0005750186,0.0001676425],"domain_scores_gemma":[0.9822954,0.0141032,0.0005157485,0.00145892,0.001177534,0.0004492982],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","study_design_scores_codex":[0.004657454,0.001575453,0.02531298,0.0009491032,0.0004923544,0.0003612762,0.001182784,0.4772616,0.007499684,0.008074931,0.01282264,0.4598098],"study_design_scores_gemma":[0.000254792,0.0007872027,0.002588404,0.00004630143,0.0001178843,0.0001025152,0.0002020441,0.9833118,0.004153419,0.006532656,0.001857313,0.0000457737],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","genre_codex":"empirical","genre_gemma":"empirical","genre_scores_codex":[0.8312286,0.001484536,0.1341675,0.001737075,0.0002478451,0.0006826969,0.002522012,0.01413453,0.01379509],"genre_scores_gemma":[0.9121385,0.0002151378,0.08093693,0.0004979093,0.00003511573,0.0002637166,0.003530429,0.0002796203,0.002102542],"genre_candidate":"empirical","genre_consensus":"empirical","teacher_disagreement_score":0.01425071,"threshold_uncertainty_score":0.03871375,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.1090399724299025,"score_gpt":0.4371863413329161,"score_spread":0.3281463689030135,"validation_status":"score_only:v0-immature-baseline","note":"Baseline scores from an immature model (maturity gate not passed). Scores rank; they never assert a category."}}