benchmark
.
space
benchmarks
rankings
compare
voices
transcripts
papers
articles
OpenAI
GPT-5.3 Codex
6 benchmarks
GPQA Diamond
#21 of 118
91.5%
SWE-bench Verified
#10 of 86
80.0%
Terminal-Bench 2.0
#6 of 29
77.3%
OSWorld
#13 of 31
64.0%
SWE-bench Pro
#21 of 43
56.8%
Humanity's Last Exam
#40 of 72
39.9%