–
weft · No library run 2
34.6Library score
Pass rate83.5%
Simplicity47.5
Problems $$4.48
Implementers get no library and write everything from scratch.
ImplementerPass rateSimplicityCostScore
DeepSeek V4.1 Flash · mini-SWE89.0%41.4$1.30
GLM 5.3 Flash · mini-SWE75.8%51.0$1.70
GPT-5.6 Luna · Codex85.8%50.0$1.48
ProblemDeepSeek V4.1 FlashGLM 5.3 FlashGPT-5.6 LunaMean
bag-rules29.225.829.028.0ninja-build30.47.227.021.5mail-addresses35.044.156.345.1property-paths39.159.865.054.7formula-ledger29.131.440.933.8record-spans27.955.043.542.1block-events48.350.050.049.4command-errors27.621.526.325.1route-tags36.011.445.030.8template-audit23.536.625.828.6grammar-inventory38.236.824.133.0grammar-trace41.139.552.144.3nesting-guard21.31.210.210.9unicode-spans30.637.840.936.4