GPT-4o System Card

Benchmark Model Rank Results
audio-understanding-reasoning-on-mmauGPT-4o Audio–Accuracy (Avg, Test-mini): 62.50
audio-understanding-reasoning-on-mmauGPT-4o mini Audio–Accuracy (Avg, Test-mini): 53.00
spatial-reasoning-on-6-dof-spatialbenchGPT-4o–Total: 36.2Position-rel: 49.4Position-abs: 28.4…
video-question-answering-on-tvbenchGPT4o 8 frames–Average Accuracy: 39.9
visual-question-answering-vqa-on-vlm2-benchGPT-4o–GC-mat: 37.45GC-trk: 39.27OC-cpr: 74.17OC-cnt: 80.62…