benchmark
.
space
benchmarks
rankings
compare
voices
transcripts
papers
articles
head to head
GPT-5.5
vs
DeepSeek V4 Pro
9 shared benchmarks
8
wins
0
ties
1
wins
38.5%
APEX-Agents
38.3%
84.4%
BrowseComp
83.4%
93.5%
GPQA Diamond
90.1%
41.4%
Humanity's Last Exam
37.7%
82.8%
MCP-Atlas
73.6%
74.0%
MRCR v2
83.5%
58.6%
SWE-bench Pro
55.4%
82.7%
Terminal-Bench 2.0
67.9%
55.6%
Toolathlon
51.8%