{"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":"codex-gemma-dda1882f352a","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0009576255,0.00009191597,0.0002413437,0.0005420116,0.0001059726,0.000006109785,0.0002108794,0.0001272143,0.00007113722],"category_scores_gemma":[0.0002672104,0.00008847501,0.00002872436,0.001092974,0.0007127207,0.0002805433,0.00005837882,0.0001905504,0.0000129005],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0001557654,"about_ca_system_score_gemma":0.0007082849,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.0006754784,"about_ca_topic_score_gemma":0.000002679225,"domain_scores_codex":[0.998698,0.00004857776,0.0004066052,0.0003415567,0.0001647381,0.0003405043],"domain_scores_gemma":[0.9992393,0.00007871415,0.0000806141,0.0002986329,0.0002361561,0.00006662204],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"observational","study_design_gemma":"observational","study_design_scores_codex":[0.0001241061,0.0001551892,0.9697983,0.0001177121,0.000004244434,0.00005497439,0.000635496,0.001577603,0.01857122,0.002155608,0.00001707895,0.006788431],"study_design_scores_gemma":[0.0001336336,0.0008212493,0.8321985,0.0004617393,0.00002037659,0.002888741,0.0009106953,0.07877713,0.07424004,0.009347681,0.0000215486,0.0001786746],"study_design_candidate":"observational","study_design_consensus":"observational","genre_codex":"empirical","genre_gemma":"empirical","genre_scores_codex":[0.9616411,0.0001955645,0.0001517332,0.0006053878,0.0008761044,0.0002346367,6.203596e-7,0.00001155822,0.03628324],"genre_scores_gemma":[0.9976006,0.00002786662,0.001478114,0.00006786415,0.00002763622,0.00002527778,0.000002210733,0.000004314448,0.0007661287],"genre_candidate":"empirical","genre_consensus":"empirical","teacher_disagreement_score":0.1375999,"threshold_uncertainty_score":0.3607906,"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."}}