LatentChem: From Textual CoT to Latent Thinking in Chemical Reasoning

Abstract

Current chemical large language models (LLMs) predominantly rely on explicit Chain-of-Thought (CoT) to solve complex reasoning problems. However, forcing nonverbal tacit chemical logic into discrete natural language imposes a fundamental ``modality mismatch,'' creating an artificial bottleneck for reasoning. To investigate this, we introduce LatentChem, a reasoning interface that decouples chemical logic from linguistic generation, enabling the model to process information via continuous thought vectors and dynamic perception. Our investigation reveals a pivotal emergent behavior: spontaneous internalization. When optimized for task success, the model voluntarily abandons verbose textual derivations in favor of implicit latent computation, suggesting that it autonomously identifies the continuous manifold as a more native substrate for chemical logic. This paradigm shift also proves to be a superior computational strategy: LatentChem achieves a 59.88\% non-tie win rate against the strong CoT baseline on the rigorous ChemCoTBench, while delivering a broad 10.84$\times$ average speedup across all evaluated benchmarks. This empirically validates that chemical logic is inherently better modeled by continuous latent dynamics than by linear linguistic approximations.