GPT-5.6 Luna
9 benchmarks
GPQA Diamond
#19 of 118
92.3%
Terminal-Bench 2.1
#8 of 27
84.7%
BrowseComp
#18 of 45
83.3%
FrontierMath
#4 of 11
78.6%
MMMU Pro
#16 of 36
78.4%
DeepSWE
#8 of 22
67.2%
SWE-bench Pro
#9 of 43
62.7%
OSWorld 2.0
#6 of 10
45.6%
ARC-AGI 3
#7 of 8
0.18%