weft · Opus 5.5 run 2
34.0Library score
Pass rate78.7%
Simplicity52.1
Problems $$10.42
Library $$12.16
The library
weft · designed by Opus 5.5 (mini-SWE)6,976 source lines
ImplementerPass rateSimplicityCostScore
DeepSeek V4.1 Flash · mini-SWE86.2%46.8$2.03
GLM 5.3 Flash · mini-SWE74.4%58.8$5.55
GPT-5.6 Luna · Codex75.4%50.9$2.85
ProblemDeepSeek V4.1 FlashGLM 5.3 FlashGPT-5.6 LunaMean
bag-rules42.120.715.826.2ninja-build6.935.133.125.0mail-addresses34.941.556.344.2property-paths47.040.222.936.7formula-ledger39.124.429.631.0record-spans46.654.752.851.4block-events49.563.045.452.6command-errors19.229.223.123.8route-tags59.139.239.545.9template-audit34.029.64.622.7grammar-inventory24.624.026.224.9grammar-trace41.671.628.447.2nesting-guard27.71.917.115.6unicode-spans31.720.033.828.5