[2603.02112] Recursive Models for Long-Horizon Reasoning

Computer Science > Machine Learning arXiv:2603.02112 (cs) [Submitted on 2 Mar 2026] Title:Recursive Models for Long-Horizon Reasoning Authors:Chenxiao Yang, Nathan Srebro, Zhiyuan Li View a PDF of the paper titled Recursive Models for Long-Horizon Reasoning, by Chenxiao Yang and 2 other authors View PDF HTML (experimental) Abstract:Modern language models reason within bounded context, an inherent constraint that poses a fundamental barrier to long-horizon reasoning. We identify recursion as a core principle for overcoming this barrier, and propose recursive models as a minimal realization, where the model can recursively invoke itself to solve subtasks in isolated contexts. We prove that any computable problem admits a recursive decomposition in which each subtask requires only exponentially smaller active context than standard autoregressive models; this strictly surpasses any context management approach confined to a single sequence, such as summarization. We further generalize our framework to modern agentic systems with arbitrary context processing and control flows, and prove that recursive models can achieve optimal power within this broader class. Experimentally, we train a 3B model to reason recursively and evaluate on Boolean satisfiability, a task requiring long-horizon combinatorial search, where it significantly outperforms frontier LLMs. Subjects: Machine Learning (cs.LG); Computation and Language (cs.CL) Cite as: arXiv:2603.02112 [cs.LG]   (or arXiv:2603.0211...