Reddit LocalLLaMA Debate: Are Frontier Models Really Getting Smarter?
AI-generated summary A viral Reddit LocalLLaMA post examines whether frontier models are actually getting smarter in core problem-solving ability. Commenters are split on the topic while a former frontier evaluator argues post-training has internalized what scaffolding used to do.