Claude Opus 4.6 vs DeepSeek V4 Pro — Who Actually Wins? Real Benchmark Data (2026)

head to head

Claude Opus 4.6 vs DeepSeek V4 Pro

11 shared benchmarks

7

wins

0

ties

4

wins

100.0%

AIME

95.2%

84.0%

BrowseComp

83.4%

91.3%

GPQA Diamond

90.1%

40.0%

Humanity's Last Exam

37.7%

76.0%

LiveCodeBench

93.5%

82.0%

MMLU Pro

87.5%

93.0%

MRCR v2

83.5%

77.8%

SWE-bench Multilingual

76.2%

53.4%

SWE-bench Pro

55.4%

81.42%

SWE-bench Verified

80.6%

65.4%

Terminal-Bench 2.0

67.9%