Continual-ARCby unify

ReplayARC-AGI-1 · Fresh · GPT-6 Luna, medium

Prime Agent, memory wiped

Everything the harness wrote deleted before every instance.

Held-out tasks, budget used
24.9 solved 93%
Model spend
$1.04 875 calls
Instances
100 3.4 hours

Budget used is the share of the cap of 16 that an instance was charged, averaged over the instances: a solved instance costs 1 for each wrong answer and each demonstration, an unsolved one all 16, so lower is better.

Loading the run