umami · Kimi K3 run 2

47.5Library score
Pass rate95.1%
Simplicity54.0
Problems $$7.67
Library $$7.69
The library
umami · designed by Kimi K3 (mini-SWE)1,470 source lines
ImplementerPass rateSimplicityCostScore
DeepSeek V4.1 Flash · mini-SWE97.7%57.1$2.23
53.6
GLM 5.3 Flash · mini-SWE94.9%54.6$4.05
45.3
GPT-5.6 Luna · Codex92.6%50.5$1.40
43.7
ProblemDeepSeek V4.1 FlashGLM 5.3 FlashGPT-5.6 LunaMean

Click a problem to see how each solution's score is built.