The Efficiency Paradox: Pathway's 150M Model Challenges the 'Think Out Loud' Orthodoxy

The prevailing belief that AI must "think out loud"—generating lengthy chains of thought to solve complex problems—is being challenged by a new class of models that reason internally. The hypothesis: intelligence per dollar can be dramatically improved by ditching verbose intermediate steps.

Evidence For

• Pathway's BDH-CQ model, a 150M-parameter post-transformer architecture, scored 29.5% pass@2 on the ARC-AGI-1 benchmark while costing just $0.0007 per task—making...