A 150M-parameter model pushes the ARC-AGI-1 cost frontier at $0.0007 per task

BDH-CQ (arXiv, submitted 10 August 2026) pairs in-context learning with recurrent latent reasoning: demonstrations update a recurrent memory at inference time, then the model iterates in a high-dimensional latent space without verbalising intermediate steps. The 150M-parameter version reports 29.5% pass@2 on ARC-AGI-1 at a computed $0.0007 per task, which the authors position as past the previously reported cost-accuracy frontier.

What it means The interesting axis is cost per solved task, not raw score — a small recurrent model competing on that axis is a different procurement argument than a bigger frontier call.