weft · Opus 5.5 run 3

38.0Library score
Pass rate86.4%
Simplicity51.4
Problems $$11.15
Library $$10.78
The library
weft · designed by Opus 5.5 (mini-SWE)6,969 source lines
ImplementerPass rateSimplicityCostScore
DeepSeek V4.1 Flash · mini-SWE91.2%49.5$1.94
41.8
GLM 5.3 Flash · mini-SWE81.7%54.8$6.37
34.4
GPT-5.6 Luna · Codex86.1%50.0$2.85
37.9
ProblemDeepSeek V4.1 FlashGLM 5.3 FlashGPT-5.6 LunaMean
bag-rules34.426.528.929.9ninja-build23.826.230.626.9mail-addresses48.051.557.852.4property-paths50.742.649.447.6formula-ledger35.418.631.028.3record-spans56.249.044.449.9block-events58.253.041.650.9command-errors51.126.725.634.4route-tags46.133.450.343.2template-audit43.434.830.936.4grammar-inventory21.30.033.018.1grammar-trace67.056.557.160.2nesting-guard22.823.316.020.7unicode-spans26.739.733.933.5