Op
gpt-5.4-nano Benchmark & Insights
OpenAI OpenAI API
Updated Apr 13, 2026 All models
Sample size
9 runs
in window
Accuracy
66.7%
consensus match · 9d
Confidence
71%
over 9 runs
Window end
Apr 13, 2026
most recent run
Input price
$0.20/MTok
prompt tokens
Output price
$1.25/MTok
completion tokens
Model insights
- 01 Undersampled at 9 days — provisional.
- 02 All 3 misses overcall, one a false "unsafe", continuing the pattern that nano-tier gpt models trade away accuracy far more steeply than the mini tier.