B
Sep 8, 2026BDH-CQ: Breaking the ARC-AGI Cost-Accuracy Frontier with Latent Reasoning
A new model, BDH-CQ, introduces recurrent latent reasoning to solve complex tasks without verbalizing...
Sep 8, 20264 min read0 reactions0 comments
Tag archive
A new model, BDH-CQ, introduces recurrent latent reasoning to solve complex tasks without verbalizing...
The top ARC-AGI-3 entry on ARC Prize's public leaderboard is an NVIDIA-labelled agent scoring 85.1% on the public demo set at a cost of $332, self-reported and not independently verified, and it is not the AVO system that viral posts credited with a
Pathway's BDH-CQ solves reasoning puzzles by iterating in a latent workspace instead of writing out its thinking, reaching 29.5% on public ARC-AGI-1 at seven hundredths of a cent per task and setting a record in cost efficiency rather than accuracy.