heretical · Kimi K3 run 3

46.5Library score
Pass rate90.3%
Simplicity57.2
Problems $$10.91
Library $$6.40
The library
heretical · designed by Kimi K3 (mini-SWE)2,817 source lines
ImplementerPass rateSimplicityCostScore
DeepSeek V4.1 Flash · mini-SWE91.9%59.2$3.76
49.5
GLM 5.3 Flash · mini-SWE87.0%57.5$4.84
43.3
GPT-5.6 Luna · Codex92.1%54.9$2.31
46.6
ProblemDeepSeek V4.1 FlashGLM 5.3 FlashGPT-5.6 LunaMean

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