umami · DeepSeek V4 Flash run 3
54.6Run score
Pass rate95.1%
Simplicity61.0
Problems $$3.28
DeepSeek V4 Flash uses the real production library (unagi-chan).
ImplementerPass rateSimplicityCostScore
DeepSeek V4 Flash · mini-SWE95.1%61.0$3.28
ProblemDeepSeek V4 Flash
Click a problem to see how each solution's score is built.