benchmark
.
space
benchmarks
rankings
compare
voices
transcripts
papers
articles
head to head
DeepSeek V4 Pro
vs
GLM-5
7 shared benchmarks
7
wins
0
ties
0
wins
95.2%
AIME
92.7%
83.4%
BrowseComp
75.9%
90.1%
GPQA Diamond
86.0%
37.7%
Humanity's Last Exam
30.5%
83.5%
MRCR v2
26.3%
80.6%
SWE-bench Verified
77.8%
67.9%
Terminal-Bench 2.0
56.2%