Claude Opus 4.7 vs Claude Opus 4.6
18 shared benchmarks
15
wins
0
ties
3
wins
75.8%
ARC-AGI 2
85.0%
79.0%
Aider Polyglot
72.0%
90.9%
BigLaw Bench
90.2%
79.3%
BrowseComp
84.0%
82.1%
CharXiv
61.5%
73.8%
CyberGym
66.6%
94.2%
GPQA Diamond
91.3%
46.9%
Humanity's Last Exam
40.0%
83.0%
LiveCodeBench
76.0%
87.0%
MMLU Pro
82.0%
92.0%
MMMLU
91.1%
59.2%
MRCR v2
93.0%
78.0%
OSWorld
72.7%
85.7%
SWE-bench Multilingual
77.8%
35.0%
SWE-bench Multimodal
27.1%
64.3%
SWE-bench Pro
53.4%
87.6%
SWE-bench Verified
81.42%
77.0%
Terminal-Bench 2.0
65.4%