# SLPO: Scaling Latent Reasoning via a Surrogate Policy

arXiv: https://arxiv.org/abs/2607.19691 · HTML: https://mlsanity.com/paper/2607.19691/

## Key facts

- No citing papers in the map yet (data through 14 Oct 2026).
- Placed in Test-time compute and efficient thinking, one of 17 directions in LLM reasoning.
- Builds on 22 of its 25 references that are in the map.
- First posted on arXiv 22 Jul 2026.

## Placed in

- LLM reasoning / [Test-time compute and efficient thinking](https://mlsanity.com/lr/test-time-compute-and-efficient-thinking/)

## Builds on (most cited first, 22 of 22)

- [Chain-of-Thought Prompting Elicits Reasoning in Large Language Models](https://mlsanity.com/paper/2201.11903/) (2022)
- [DeepSeekMath: Pushing the Limits of Mathematical Reasoning in Open Language Models](https://mlsanity.com/paper/2402.03300/) (2024)
- [Self-Consistency Improves Chain of Thought Reasoning in Language Models](https://mlsanity.com/paper/2203.11171/) (2022)
- [s1: Simple test-time scaling](https://mlsanity.com/paper/2501.19393/) (2025)
- [Does Reinforcement Learning Really Incentivize Reasoning Capacity in LLMs Beyond the Base Model?](https://mlsanity.com/paper/2504.13837/) (2025)
- [Training Large Language Models to Reason in a Continuous Latent Space](https://mlsanity.com/paper/2412.06769/) (2024)
- [Stop Overthinking: A Survey on Efficient Reasoning for Large Language Models](https://mlsanity.com/paper/2503.16419/) (2025)
- [Reinforcement Learning with Verifiable Rewards Implicitly Incentivizes Correct Reasoning in Base LLMs](https://mlsanity.com/paper/2506.14245/) (2025)
- [CODI: Compressing Chain-of-Thought into Continuous Space via Self-Distillation](https://mlsanity.com/paper/2502.21074/) (2025)
- [A Survey of Reinforcement Learning for Large Reasoning Models](https://mlsanity.com/paper/2509.08827/) (2025)
- [Soft Thinking: Unlocking the Reasoning Potential of LLMs in Continuous Concept Space](https://mlsanity.com/paper/2505.15778/) (2025)
- [Think Silently, Think Fast: Dynamic Latent Compression of LLM Reasoning Chains](https://mlsanity.com/paper/2505.16552/) (2025)
- [Reasoning Beyond Language: A Comprehensive Survey on Latent Chain-of-Thought Reasoning](https://mlsanity.com/paper/2505.16782/) (2025)
- [SIM-CoT: Supervised Implicit Chain-of-Thought](https://mlsanity.com/paper/2509.20317/) (2025)
- [LLM Latent Reasoning as Chain of Superposition](https://mlsanity.com/paper/2510.15522/) (2025)
- [Render-of-Thought: Rendering Textual Chain-of-Thought as Images for Visual Latent Reasoning](https://mlsanity.com/paper/2601.14750/) (2026)
- [The Reasoning Boundary Paradox: How Reinforcement Learning Constrains Language Models](https://mlsanity.com/paper/2510.02230/) (2025)
- [Parallel Test-Time Scaling for Latent Reasoning Models](https://mlsanity.com/paper/2510.07745/) (2025)
- [ReGuLaR: Variational Latent Reasoning Guided by Rendered Chain-of-Thought](https://mlsanity.com/paper/2601.23184/) (2026)
- [Latent-GRPO: Group Relative Policy Optimization for Latent Reasoning](https://mlsanity.com/paper/2604.27998/) (2026)
- [DART: Distilling Autoregressive Reasoning to Silent Thought](https://mlsanity.com/paper/2506.11752/) (2025)
- [Effective Reinforcement Learning for Reasoning in Language Models](https://mlsanity.com/paper/2505.17218/) (2025)

## Abstract (arXiv)

Reinforcement learning with verifiable rewards has become the predominant recipe for eliciting test-time scaling in explicit Chain-of-Thought reasoners. Yet this scaling path remains computationally costly, since intermediate reasoning must be externalized as natural-language tokens. Latent reasoning instead carries intermediate computation as continuous vectors and already matches or surpasses explicit CoT at far shorter horizons. Despite this promise, latent reasoners remain largely imitation-bound, while explicit CoT has already moved past imitation via outcome-reward RL. Latent trajectories lack a tractable per-step likelihood and an adaptive stopping interface under fixed thinking budgets, so outcome rewards cannot elicit latent test-time scaling. We introduce Surrogate Latent Policy Optimization (SLPO) to bring outcome-reward RL to autoregressive latent reasoners: a differentiable surrogate policy interface over latent transitions for trajectory-level credit assignment, and a correctness-supervised stopping head that outcome-reward optimization refines into a variable-horizon policy. Across two continuous latent reasoners, two backbones, and three held-out benchmarks, SLPO improves Pass@8 and Pass@16 in all 12 backbone--dataset settings, with gains of up to 12.07 percentage points. SLPO further transfers to soft-token inference and learns difficulty-adaptive computation, allocating longer latent trajectories to harder instances. Project Page: https://modalitydance.github.io/SLPO/