umami · Kimi K3 run 3

51.3Library score
Pass rate97.4%
Simplicity55.1
Problems $$6.86
Library $$4.35
The library
umami · designed by Kimi K3 (mini-SWE)1,219 source lines
ImplementerPass rateSimplicityCostScore
DeepSeek V4.1 Flash · mini-SWE97.6%58.5$1.84
54.9
GLM 5.3 Flash · mini-SWE98.2%55.6$3.61
52.1
GPT-5.6 Luna · Codex96.3%51.3$1.41
46.8
ProblemDeepSeek V4.1 FlashGLM 5.3 FlashGPT-5.6 LunaMean

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