weft · DeepSeek V4 Pro run 2

31.3Library score
Pass rate82.7%
Simplicity42.3
Problems $$9.07
Library $$0.38
The library
weft · designed by DeepSeek V4 Pro (mini-SWE)2,514 source lines
ImplementerPass rateSimplicityCostScore
DeepSeek V4.1 Flash · mini-SWE89.2%40.6$3.34
33.7
GLM 5.3 Flash · mini-SWE72.6%43.2$3.58
27.4
GPT-5.6 Luna · Codex86.3%43.0$2.14
32.8
ProblemDeepSeek V4.1 FlashGLM 5.3 FlashGPT-5.6 LunaMean
bag-rules31.037.027.831.9ninja-build6.126.535.122.5mail-addresses45.139.443.242.6property-paths56.360.351.155.9formula-ledger37.134.634.735.5record-spans35.436.231.234.3block-events32.332.640.035.0command-errors25.923.123.024.0route-tags49.621.937.036.2template-audit34.819.321.825.3grammar-inventory25.213.527.021.9grammar-trace43.438.548.343.4nesting-guard20.60.98.19.9unicode-spans29.00.030.619.9