Claude Opus 4.8 vs Claude Sonnet 4.6 — Who Actually Wins? Real Benchmark Data (2026)

head to head

Claude Opus 4.8 vs Claude Sonnet 4.6

9 shared benchmarks

9

wins

0

ties

0

wins

95.7%

AIME

94.0%

84.3%

BrowseComp

74.0%

80.5%

CharXiv

72.4%

59.0%

DeepSWE

30.0%

91.0%

GPQA Diamond

89.9%

49.8%

Humanity's Last Exam

33.2%

77.8%

MCP-Atlas

69.5%

78.9%

MMMU Pro

74.5%

54.8%

OSWorld 2.0

41.5%