umami · GPT-6 Luna run 1
48.9Run score
Pass rate90.4%
Simplicity67.4
Problems $$0.18
GPT-6 Luna uses the real production library (unagi-chan).
ImplementerPass rateSimplicityCostScore
GPT-6 Luna · mini-SWE90.4%67.4$0.18
ProblemGPT-6 Luna
Click a problem to see how each solution's score is built.