roadkill · Kimi K3 run 3

16.0Library score
Pass rate68.9%
Simplicity33.0
Problems $$8.06
Library $$4.46
The library
roadkill · designed by Kimi K3 (mini-SWE)2,104 source lines
ImplementerPass rateSimplicityCostScore
DeepSeek V4.1 Flash · mini-SWE70.1%34.1$1.94
18.3
GLM 5.3 Flash · mini-SWE67.5%28.7$4.52
11.0
GPT-5.6 Luna · Codex68.8%35.6$1.60
18.7
ProblemDeepSeek V4.1 FlashGLM 5.3 FlashGPT-5.6 LunaMean
transit-route-check1.78.81.84.1live-session5.013.126.214.8charging-routes17.60.00.05.9link-time-bins25.820.221.422.5explicit-fleet11.712.815.613.4braess-network19.26.623.716.5signal-plan-sweep12.94.218.311.8merge-spillback4.27.84.65.5incident-guidance32.732.237.534.1corridor-replications25.52.028.618.7speed-limit-control6.30.016.67.6reactive-signals47.216.338.834.1signal-progression15.57.813.012.1gridlock-gating18.48.320.715.8workzone-recovery3.78.911.98.2district-los45.627.520.531.2