arazu · GPT-6 Sol run 3

36.4Library score
Pass rate72.0%
Simplicity62.6
Problems $$12.30
Library $$0.35
The library
arazu · designed by GPT-6 Sol (mini-SWE)2,055 source lines
ImplementerPass rateSimplicityCostScore
DeepSeek V4.1 Flash · mini-SWE70.5%59.3$3.90
37.1
GLM 5.3 Flash · mini-SWE75.1%64.0$5.59
39.0
GPT-5.6 Luna · Codex70.5%64.7$2.80
33.2
ProblemDeepSeek V4.1 FlashGLM 5.3 FlashGPT-5.6 LunaMean
lander-lidar52.451.151.851.8scene-contact28.739.237.135.0world-contact90.516.80.035.8aircraft-raycast0.062.254.738.9capsule-training11.523.318.517.8physics-tick5.09.38.37.6offset-strike43.127.443.137.9layered-laser95.994.194.995.0swept-cargo1.61.720.88.0compound-scanner32.853.140.742.2sensor-transit-ledger39.948.248.645.6articulated-creature0.40.40.50.4cable-replay0.045.918.821.6damping53.144.41.633.0one-way40.335.010.628.7material-braking32.161.936.243.4platformer-ccd46.756.445.849.6grab-replay58.550.658.355.8keyed-platform-drop52.161.533.248.9ccd-wall35.715.241.030.6motor-target27.923.20.017.0rope-cut68.937.765.857.5