arxiv
GPT-5-mini + CodeStruct
1 benchmarks
SWE-bench Verified
#53 of 72
62.0%
benchmark.space canonical public ledger of AI capability