Updated
Updated · Science News Magazine · Sep 22
BDH-CQ Solves Nearly 30% of ARC Puzzles at $0.00070 Each Without Written Reasoning
Updated
Updated · Science News Magazine · Sep 22

BDH-CQ Solves Nearly 30% of ARC Puzzles at $0.00070 Each Without Written Reasoning

1 articles · Updated · Science News Magazine · Sep 22

Summary

  • Nearly 3 in 10 ARC-AGI-1 puzzles were solved by experimental model BDH-CQ in two attempts, with the system reasoning internally instead of generating chain-of-thought text.
  • A fixed-size memory lets BDH-CQ absorb example puzzles without revisiting all prior context, aiming to cut the token-heavy computation that makes many AI reasoning systems slower and costlier.
  • At about $0.00070 per puzzle, the researchers estimate BDH-CQ runs at roughly one-eleventh the cost of GPT-5.6 Luna on the same benchmark, though the cost calculations were not directly comparable.
  • ARC-style tests ask models to infer visual rules from examples; BDH-CQ handled some shape rotations and movements better than color changes, ordering tasks and nested rule combinations.
  • Researchers and outside experts called the result promising but preliminary: the paper is unreviewed, the model is specialized for ARC problems, and hidden reasoning makes its decisions harder to inspect.

Insights

Could abandoning step-by-step text explanations be the ultimate secret to making AI reasoning exponentially cheaper and faster?
If AI no longer needs to think aloud in words, what hidden logic is driving its decisions in the dark?