Platform benchmark

claude-opus-5

anthropic · GPQA-Diamond · seed 42 · 2026-07-24 22:50 UTC

ResultsGPQA-Diamond · n=96 · seed 42
93.8%

93.8% correct · 99.0% answered on GPQA-Diamond

Score per $
0.23
Run cost
$4.00
Latency p50
11.8 s
Latency p95
56.4 s
Items
96/96 completed
Dropped
0