P

roadkill · Production library run 3

28.7Library score
Pass rate75.1%
Simplicity45.9
Problems $$13.12

Implementers use the real production library (uxsim).

ImplementerPass rateSimplicityCostScore
DeepSeek V4.1 Flash · mini-SWE84.5%48.8$1.89
35.4
GLM 5.3 Flash · mini-SWE79.1%40.4$9.03
26.5
GPT-5.6 Luna · Codex61.8%48.6$2.21
24.3
ProblemDeepSeek V4.1 FlashGLM 5.3 FlashGPT-5.6 LunaMean
transit-route-check31.444.017.831.1live-session7.47.937.017.4charging-routes67.046.01.938.3link-time-bins80.561.170.570.7explicit-fleet45.115.238.432.9braess-network38.912.832.528.1signal-plan-sweep5.510.036.817.4merge-spillback38.912.013.821.6incident-guidance16.734.20.217.1corridor-replications23.436.60.020.0speed-limit-control15.718.831.522.0reactive-signals50.045.330.441.9signal-progression28.831.339.933.3gridlock-gating41.434.235.937.2workzone-recovery61.52.71.822.0district-los13.411.30.98.5