Claude Opus 4.6 vs Claude Opus 4.8
6 shared benchmarks
2
wins
0
ties
4
wins
100.0%
AIME
95.7%
84.0%
BrowseComp
84.3%
61.5%
CharXiv
80.5%
91.3%
GPQA Diamond
91.0%
40.0%
Humanity's Last Exam
49.8%
53.4%
SWE-bench Pro
69.2%