weft · Grok 4.6 run 2
33.4Library score
Pass rate83.5%
Simplicity45.6
Problems $$11.65
Library $$2.74
The library
weft · designed by Grok 4.6 (mini-SWE)5,213 source lines
ImplementerPass rateSimplicityCostScore
DeepSeek V4.1 Flash · mini-SWE89.6%43.7$3.49
GLM 5.3 Flash · mini-SWE75.1%47.7$5.62
GPT-5.6 Luna · Codex85.8%45.5$2.54
ProblemDeepSeek V4.1 FlashGLM 5.3 FlashGPT-5.6 LunaMean
bag-rules27.918.432.326.2ninja-build32.327.823.227.8mail-addresses41.736.748.142.2property-paths27.640.943.637.4formula-ledger43.132.927.634.5record-spans47.648.040.745.4block-events45.172.839.052.3command-errors21.521.624.922.7route-tags40.80.145.428.8template-audit34.239.340.037.8grammar-inventory27.824.119.623.8grammar-trace51.240.042.044.4nesting-guard23.21.417.814.1unicode-spans27.831.732.230.6