–

arazu · No library run 2

21.9Library score
Pass rate74.2%
Simplicity42.1
Problems $$12.46

Implementers get no library and write everything from scratch.

ImplementerPass rateSimplicityCostScore
DeepSeek V4.1 Flash · mini-SWE77.2%36.5$1.81
23.3
GLM 5.3 Flash · mini-SWE68.5%44.2$7.99
18.1
GPT-5.6 Luna · Codex76.2%46.0$2.66
24.3
ProblemDeepSeek V4.1 FlashGLM 5.3 FlashGPT-5.6 LunaMean
lander-lidar30.732.926.229.9scene-contact9.314.019.714.3world-contact14.831.534.126.8aircraft-raycast22.40.031.217.8capsule-training16.513.211.113.6physics-tick7.00.04.53.8offset-strike17.60.00.46.0layered-laser30.023.927.427.1swept-cargo18.30.012.810.4compound-scanner26.723.919.623.4sensor-transit-ledger35.334.946.639.0articulated-creature0.210.663.924.9cable-replay0.00.025.78.6damping29.131.726.429.1one-way29.429.318.525.7material-braking21.60.024.915.5platformer-ccd40.339.40.026.6grab-replay36.636.726.933.4keyed-platform-drop43.528.625.832.6ccd-wall22.67.311.313.7motor-target31.49.536.325.7rope-cut30.030.541.634.0