Replay / ARC-AGI-1 · Sequential · GPT-6 Luna, low

Demo-cache floor

The model alone, shown every demonstration it bought for the task again for free.

Stream, budget used

38.8 solved 62%

Epilogue, budget used

49.0 solved 48%

Model spend

$1.52 1,574 calls

Instances

225 6 hours

Budget used is the share of the cap of 16 that an instance used, averaged over the instances: a wrong answer and a demonstration each cost 1, so lower is better.