weft · Opus 5.5 run 1
33.7Library score
Pass rate79.9%
Simplicity53.4
Problems $$10.25
Library $$11.11
The library
weft · designed by Opus 5.5 (mini-SWE)6,332 source lines
ImplementerPass rateSimplicityCostScore
DeepSeek V4.1 Flash · mini-SWE83.9%49.1$2.10
GLM 5.3 Flash · mini-SWE68.7%61.7$5.62
GPT-5.6 Luna · Codex84.8%51.1$2.53
ProblemDeepSeek V4.1 FlashGLM 5.3 FlashGPT-5.6 LunaMean
bag-rules22.723.239.628.5ninja-build29.80.033.721.1mail-addresses44.00.043.629.2property-paths35.40.042.225.9formula-ledger31.144.541.539.0record-spans52.431.629.137.7block-events35.857.260.551.2command-errors25.744.652.040.8route-tags69.80.139.836.6template-audit41.843.731.939.1grammar-inventory32.338.525.232.0grammar-trace69.461.847.859.7nesting-guard2.02.98.24.4unicode-spans29.918.533.427.3