Claude Opus 4.6 vs Step-3.5-Flash
7 shared benchmarks
5
wins
0
ties
2
wins
100.0%
AIME
97.3%
84.0%
BrowseComp
69.0%
91.3%
GPQA Diamond
83.5%
76.0%
LiveCodeBench
86.4%
82.0%
MMLU Pro
84.4%
81.42%
SWE-bench Verified
74.4%
81.8%
Terminal-Bench 2.0
51.0%