Opus 5.5

66.9± 7.1 (95% CI)

mini-SWE · pass rate 93.0% · simplicity 78.3

Runs

Each cell is one full attempt: Opus 5.5 solves every problem with the task's production library. Click a cell to open that run.

TaskRun 1Run 2Run 3Mean ± std

arazu 68.9 67.0 65.667.2 ± 1.6

clirs 72.0 83.3 82.379.2 ± 6.2

heretical 44.8 45.5 47.545.9 ± 1.4

npr 62.7 63.8 63.963.4 ± 0.7

plaid 80.5 83.2 81.381.7 ± 1.4

pyda 73.7 70.3 74.772.9 ± 2.3

canon 80.1 82.1 80.981.0 ± 1.0

roadkill 40.1 30.9 38.936.6 ± 5.0

rubberband 74.1 71.8 70.872.2 ± 1.7

weft 71.3 71.9 73.172.1 ± 0.9

sapi 50.8 55.4 52.452.9 ± 2.3

simu 65.4 67.0 58.163.5 ± 4.8

uglypie 74.8 75.1 74.274.7 ± 0.5

umami 83.8 80.4 81.081.7 ± 1.8

wapi 63.8 62.1 59.961.9 ± 2.0