Op
gpt-6-luna
Benchmark & Insights
OpenAI OpenAI API
Updated Oct 4, 2026 All models
Sample size
232 runs
in window
Accuracy
74.1%
reference match · 282d
Confidence
85%
over 282 runs
Window end
Oct 4, 2026
most recent run
Input price
$0.10/MTok
prompt tokens
Output price
$0.50/MTok
completion tokens
Model insights
- 01 The cheapest OpenAI model ($0.10/$0.50) is mid-pack, matching gpt-5.6-luna at half its price.
- 02 Its weak spot is false alarms: it rated 12 "MEDIUM_SAFE" days "high" and "unsafe".
- 03 Started badly (40% in February) but reached 87% in August.
Notes
Twice cheaper than gpt-5.6-luna. More about GPT-6 Luna release
Recent forecasts