benchmark
.
space
benchmarks
rankings
compare
voices
transcripts
papers
articles
head to head
GPT-5.6 Sol
vs
GPT-5.4
9 shared benchmarks
9
wins
0
ties
0
wins
7.78%
ARC-AGI 3
0.26%
52.7%
Agents' Last Exam
44.7%
90.4%
BrowseComp
82.7%
72.7%
DeepSWE
52.0%
83.0%
FrontierMath
47.6%
94.6%
GPQA Diamond
92.8%
44.5%
Humanity's Last Exam
36.24%
83.0%
MMMU Pro
81.2%
64.6%
SWE-bench Pro
57.7%