Platform benchmark

gpt-5.4-mini

openai · GPQA-Diamond · seed 42 · 2026-07-15 19:00 UTC

ResultsGPQA-Diamond · n=96 · seed 42
69.8%

69.8% correct · 100.0% answered on GPQA-Diamond

Score per $
4.03
Run cost
$0.17
Latency p50
3.0 s
Latency p95
6.3 s
Items
96/96 completed
Dropped
0