heretical · Kimi K3 run 2

42.4Library score
Pass rate86.0%
Simplicity56.8
Problems $$11.32
Library $$5.38
The library
heretical · designed by Kimi K3 (mini-SWE)2,777 source lines
ImplementerPass rateSimplicityCostScore
DeepSeek V4.1 Flash · mini-SWE87.2%57.3$3.90
43.8
GLM 5.3 Flash · mini-SWE84.1%56.6$4.94
40.3
GPT-5.6 Luna · Codex86.8%56.6$2.48
43.1
ProblemDeepSeek V4.1 FlashGLM 5.3 FlashGPT-5.6 LunaMean

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