Claude Opus 4.6 vs Gemma 4 26B A4B
11 shared benchmarks
9
wins
0
ties
2
wins
100.0%
AIME
88.3%
91.3%
GPQA Diamond
82.3%
40.0%
Humanity's Last Exam
17.2%
76.0%
LiveCodeBench
77.1%
82.0%
MMLU Pro
82.6%
91.1%
MMMLU
86.3%
93.0%
MRCR v2
44.1%
77.8%
SWE-bench Multilingual
17.3%
53.4%
SWE-bench Pro
13.8%
81.42%
SWE-bench Verified
17.4%
65.4%
Terminal-Bench 2.0
34.2%