The Depth Ceiling: On the Limits of Large Language Models in Discovering Latent Planning
Yi Xu · Philipp Jettkant · Laura Ruis
We find that language models can only discover shallow latent planning strategies on their own, but can generalize to deeper reasoning once a shorter version is learned, suggesting a temporary safety margin for CoT monitoring.
摘要
Chain-of-thought (CoT) monitoring is most informative when models are \emph{unable} to reason effectively in their latent representations without externalizing that reasoning. Yet little is known about the limits of such latent reasoning in LLMs. We test these limits by studying whether models can discover multi-step planning strategies without supervision on intermediate steps and execute them latently, within a single forward pass. Using graph path-finding tasks that precisely control the number of required latent planning steps, we uncover a striking limitation unresolved by massive scaling: tiny transformers trained from scratch discover strategies requiring up to three latent steps, fine-tuned GPT-4o and Qwen3-32B reach five, and GPT-5.4 attains seven under few-shot prompting. Although the maximum latent planning depth models can learn during training is five, the discovered strategy generalizes up to eight latent steps at test-time. This reveals a dissociation between the ability to \emph{discover} a latent strategy under final-answer supervision alone and the ability to \emph{execute} it once discovered. If similar limits hold more broadly, strategies requiring multiple coordinated latent planning steps may need to be explicitly taught or externalized.