Gemini 3.1 Pro vs Claude Opus 4.6 — Who Actually Wins? Real Benchmark Data (2026)

head to head

Gemini 3.1 Pro vs Claude Opus 4.6

14 shared benchmarks

8

wins

1

ties

5

wins

98.13%

AIME

100.0%

77.1%

ARC-AGI 2

85.0%

0.37%

ARC-AGI 3

0.5%

85.9%

BrowseComp

84.0%

83.3%

CharXiv

61.5%

94.3%

GPQA Diamond

91.3%

44.4%

Humanity's Last Exam

40.0%

90.99%

MMLU Pro

82.0%

84.9%

MRCR v2

93.0%

72.7%

OSWorld

72.7%

54.2%

SWE-bench Pro

53.4%

80.6%

SWE-bench Verified

81.42%

68.5%

Terminal-Bench 2.0

65.4%

74.4%

USAMO 2026

42.3%