Creator
Alibaba
Released
2025-09-22
Intelligence
10.7
Artificial Analysis Index
Coding
7.2
Artificial Analysis Index
In $/1M
$0.25
input tokens
Out $/1M
$0.97
output tokens
Blended $/1M
$0.43
3:1 blended
Speed
109
tokens / sec
Benchmark breakdown
Independent evaluation scores, normalized to 0–100.
GPQA Diamond
62
Humanity’s Last Exam
5
MMLU-Pro
73
SciCode
19
LiveCodeBench
42
MATH-500—
AIME 2025
52
τ²-Bench (agentic)
16
Terminal-Bench Hard
2
IFBench
31