Creator
Alibaba
Released
2026-03-02
Intelligence
14.7
Artificial Analysis Index
Coding
4.9
Artificial Analysis Index
In $/1M
$0.02
input tokens
Out $/1M
$0.10
output tokens
Blended $/1M
$0.04
3:1 blended
Speed
339
tokens / sec
Benchmark breakdown
Independent evaluation scores, normalized to 0–100.
GPQA Diamond
44
Humanity’s Last Exam
5
MMLU-Pro—
SciCode
7
LiveCodeBench—
MATH-500—
AIME 2025—
τ²-Bench (agentic)
82
Terminal-Bench Hard
4
IFBench
29