Op
gpt-5.4 Benchmark & Insights
OpenAI OpenAI API
Updated Jun 6, 2026 All models
Sample size
63 runs
in window
Accuracy
84.6%
consensus match · 65d
Confidence
91%
over 65 runs
Window end
Jun 6, 2026
most recent run
Input price
$2.50/MTok
prompt tokens
Output price
$15.00/MTok
completion tokens
Model insights
- 01 It fixed gpt-5.1's optimism by overcorrecting — all 10 misses raise risk or flip to "unsafe" on "safe" days, concentrated in April–May.
- 02 Errs on the safe side, but gpt-5.5 gets the balance right for double the price.
Notes
Breakpoint pricing at 272K tokens (doubles to $5/$22.50 above threshold)