–
weft · No library run 3
32.7Library score
Pass rate82.8%
Simplicity46.1
Problems $$4.97
Implementers get no library and write everything from scratch.
ImplementerPass rateSimplicityCostScore
DeepSeek V4.1 Flash · mini-SWE92.6%42.4$1.38
GLM 5.3 Flash · mini-SWE76.9%48.6$1.93
GPT-5.6 Luna · Codex78.7%47.4$1.66
ProblemDeepSeek V4.1 FlashGLM 5.3 FlashGPT-5.6 LunaMean
bag-rules29.338.520.929.5ninja-build25.124.233.227.5mail-addresses44.044.643.444.0property-paths55.248.150.151.1formula-ledger35.131.432.633.0record-spans47.441.77.032.0block-events40.244.650.645.2command-errors29.611.625.822.3route-tags38.10.130.522.9template-audit35.820.225.627.2grammar-inventory31.030.333.031.4grammar-trace48.940.039.942.9nesting-guard31.423.18.421.0unicode-spans22.130.931.728.2