Platform benchmark

gpt-5.4

openai · GPQA-Diamond · seed 42 · 2026-07-15 18:59 UTC

ResultsGPQA-Diamond · n=96 · seed 42
79.2%

79.2% correct · 100.0% answered on GPQA-Diamond

Score per $
0.95
Run cost
$0.83
Latency p50
5.7 s
Latency p95
12.7 s
Items
96/96 completed
Dropped
0