Creator
OpenAI
Released
2025-08-07
Intelligence
20.7
Artificial Analysis Index
Coding
21.9
Artificial Analysis Index
In $/1M
$0.25
input tokens
Out $/1M
$2.00
output tokens
Blended $/1M
$0.69
3:1 blended
Speed
98
tokens / sec
Benchmark breakdown
Independent evaluation scores, normalized to 0–100.
GPQA Diamond
69
Humanity’s Last Exam
5
MMLU-Pro
78
SciCode
37
LiveCodeBench
55
MATH-500—
AIME 2025
47
τ²-Bench (agentic)
32
Terminal-Bench Hard
14
IFBench
46