Fable 5.1

39.9± 2.5 (95% CI)

Claude Code · pass rate 84.9% · simplicity 58.3 · $14.39 per library

Runs

Each cell is one full run: the model designs the library, then 3 implementers solve every problem with it. Click a cell to open that run.

TaskRun 1Run 2Run 3Mean ± std

arazu 39.4 39.8 15.031.4 ± 14.2

clirs 57.7 57.5 65.560.2 ± 4.6

heretical 45.2 47.2 42.444.9 ± 2.5

npr 51.3 50.4 52.451.4 ± 1.0

plaid 46.2 47.3 48.047.2 ± 0.9

pyda 57.6 57.1 37.250.6 ± 11.6

canon 49.3 58.4 29.245.6 ± 14.9

roadkill 12.3 11.2 11.311.6 ± 0.6

rubberband 74.3 71.0 72.472.6 ± 1.7

weft 37.5 33.6 36.635.9 ± 2.0

sapi 19.4 41.0 39.533.3 ± 12.1

simu 38.1 35.6 36.036.6 ± 1.3

uglypie 21.8 40.1 21.127.7 ± 10.8

umami 30.1 30.5 29.229.9 ± 0.7

wapi 19.7 20.3 17.819.3 ± 1.3