sapi · Kimi K3 run 2

28.5Library score
Pass rate90.3%
Simplicity34.7
Problems $$3.33
Library $$6.14
The library
sapi · designed by Kimi K3 (mini-SWE)3,806 source lines
ImplementerPass rateSimplicityCostScore
DeepSeek V4.1 Flash · mini-SWE93.9%33.2$0.82
29.1
GLM 5.3 Flash · mini-SWE89.8%38.1$1.55
30.7
GPT-5.6 Luna · Codex87.3%32.6$0.97
25.8
ProblemDeepSeek V4.1 FlashGLM 5.3 FlashGPT-5.6 LunaMean

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