OpenAI
GPT-5.4 Nano
4 benchmarks
GPQA Diamond
#32 of 49
82.8%
SWE-bench Pro
#8 of 13
52.4%
Terminal-Bench 2.0
#14 of 14
46.3%
OSWorld
#16 of 16
39.0%
benchmark.space canonical public ledger of AI capability