–
plaid · No library run 3
36.0Library score
Pass rate95.9%
Simplicity41.8
Problems $$7.60
Implementers get no library and write everything from scratch.
ImplementerPass rateSimplicityCostScore
DeepSeek V4.1 Flash · mini-SWE97.1%40.8$1.40
GLM 5.3 Flash · mini-SWE95.3%43.0$4.40
GPT-5.6 Luna · Codex95.1%41.8$1.80
ProblemDeepSeek V4.1 FlashGLM 5.3 FlashGPT-5.6 LunaMean
ledger-balance40.043.135.939.6typed-settings44.60.024.222.9keymap-resolver32.436.132.733.7template-trace47.10.042.629.9recipe-graph50.20.052.834.3scoped-eval41.745.549.045.4temporal-normalizer32.738.137.836.2script-recovery43.632.735.937.4stack-audit37.034.134.835.3query-resolution46.926.545.239.5selection-flattener38.344.837.040.0schema-resolver38.846.330.338.5grammar-crossref39.548.939.642.7config-digest15.229.431.125.2shape-atlas37.444.236.939.5