rubberband · Kimi K3 run 3

67.6Library score
Pass rate89.7%
Simplicity79.4
Problems $$7.63
Library $$5.22
The library
rubberband · designed by Kimi K3 (mini-SWE)4,712 source lines
ImplementerPass rateSimplicityCostScore
DeepSeek V4.1 Flash · mini-SWE88.5%79.2$1.71
68.1
GLM 5.3 Flash · mini-SWE95.1%79.1$4.02
70.3
GPT-5.6 Luna · Codex85.4%79.9$1.91
64.3
ProblemDeepSeek V4.1 FlashGLM 5.3 FlashGPT-5.6 LunaMean
case-law-export77.241.556.258.3session-recall57.559.566.061.0document-finder68.879.977.075.2code-chunk-search62.875.249.262.4symbol-resolution46.889.354.963.7docs-search90.384.788.287.7knowledge-ingest100.079.692.490.7federated-search84.787.867.880.1post-indexer91.290.799.993.9blog-rebuild0.958.50.920.1mailbox-tags91.081.995.989.6attachment-explorer86.875.978.480.4archive-search97.156.591.681.7index-admin4.239.72.115.3file-locator75.978.977.577.4project-rule-search90.152.380.374.2blog-filters58.961.336.752.3dictionary-lookup45.558.338.347.4archive-timeline45.470.937.251.2source-rule-search86.983.596.388.9