Gemma 4 31B
16 benchmarks
AIME
#46 of 117
89.2%
MMMLU
#3 of 7
88.4%
τ2-bench
#6 of 9
86.4%
MATH-Vision
#6 of 15
85.6%
MMLU Pro
#15 of 55
85.2%
GPQA Diamond
#57 of 118
84.3%
LiveCodeBench
#23 of 53
80.0%
MMMU Pro
#17 of 36
76.9%
MRCR v2
#10 of 18
66.4%
MedXPertQA MM
#1 of 4
61.3%
SWE-bench Verified
#75 of 86
52.0%
SWE-bench Multilingual
#14 of 16
51.7%
Terminal-Bench 2.0
#27 of 29
42.9%
SWE-bench Pro
#42 of 43
35.7%
Humanity's Last Exam
#56 of 72
26.5%
Humanity's Last Exam
#56 of 72
19.5%