–

arazu · No library run 3

23.0Library score
Pass rate75.5%
Simplicity42.0
Problems $$11.04

Implementers get no library and write everything from scratch.

ImplementerPass rateSimplicityCostScore
DeepSeek V4.1 Flash · mini-SWE82.4%34.7$1.63
23.1
GLM 5.3 Flash · mini-SWE65.6%49.9$6.46
20.3
GPT-5.6 Luna · Codex77.1%42.5$2.95
25.5
ProblemDeepSeek V4.1 FlashGLM 5.3 FlashGPT-5.6 LunaMean
lander-lidar23.40.036.319.9scene-contact21.30.024.515.3world-contact36.248.731.038.6aircraft-raycast25.60.028.618.1capsule-training12.123.214.916.8physics-tick6.80.011.56.1offset-strike13.218.37.513.0layered-laser28.432.026.228.9swept-cargo15.20.014.69.9compound-scanner23.724.313.420.4sensor-transit-ledger34.631.141.535.8articulated-creature29.90.347.625.9cable-replay0.05.60.01.9damping24.534.724.828.0one-way24.20.045.823.3material-braking25.60.018.514.7platformer-ccd35.242.133.436.9grab-replay31.419.625.225.4keyed-platform-drop30.841.727.433.3ccd-wall15.838.042.432.0motor-target24.731.40.018.7rope-cut25.455.846.542.6