# OPLD: On-Policy Latent Distillation for Multimodal Reasoning

arXiv: https://arxiv.org/abs/2607.28154 · HTML: https://mlsanity.com/paper/2607.28154/

## Key facts

- No citing papers in the map yet (data through 14 Oct 2026).
- Placed in Multimodal reasoning, one of 17 directions in LLM reasoning.
- Builds on 21 of its 33 references that are in the map.
- First posted on arXiv 30 Jul 2026.

## Placed in

- LLM reasoning / [Multimodal reasoning](https://mlsanity.com/lr/multimodal-reasoning/)

## Builds on (most cited first, 21 of 21)

- [Training Large Language Models to Reason in a Continuous Latent Space](https://mlsanity.com/paper/2412.06769/) (2024)
- [Multimodal Chain-of-Thought Reasoning in Language Models](https://mlsanity.com/paper/2302.00923/) (2023)
- [DeepEyes: Incentivizing "Thinking with Images" via Reinforcement Learning](https://mlsanity.com/paper/2505.14362/) (2025)
- [Visual CoT: Advancing Multi-Modal Language Models with a Comprehensive Dataset and Benchmark for Chain-of-Thought Reasoning](https://mlsanity.com/paper/2403.16999/) (2024)
- [Machine Mental Imagery: Empower Multimodal Reasoning with Latent Visual Tokens](https://mlsanity.com/paper/2506.17218/) (2025)
- [Latent Visual Reasoning](https://mlsanity.com/paper/2509.24251/) (2025)
- [Adaptive Chain-of-Focus Reasoning via Dynamic Visual Search and Zooming for Efficient VLMs](https://mlsanity.com/paper/2505.15436/) (2025)
- [Zebra-CoT: A Dataset for Interleaved Vision Language Reasoning](https://mlsanity.com/paper/2507.16746/) (2025)
- [Thinking with Generated Images](https://mlsanity.com/paper/2505.22525/) (2025)
- [VGR: Visual Grounded Reasoning](https://mlsanity.com/paper/2506.11991/) (2025)
- [Image-of-Thought Prompting for Visual Reasoning Refinement in Multimodal Large Language Models](https://mlsanity.com/paper/2405.13872/) (2024)
- [Grounded Chain-of-Thought for Multimodal Large Language Models](https://mlsanity.com/paper/2503.12799/) (2025)
- [Sketch-in-Latents: Eliciting Unified Reasoning in MLLMs](https://mlsanity.com/paper/2512.16584/) (2025)
- [Forest Before Trees: Latent Superposition for Efficient Visual Reasoning](https://mlsanity.com/paper/2601.06803/) (2026)
- [Vision-aligned Latent Reasoning for Multi-modal Large Language Model](https://mlsanity.com/paper/2602.04476/) (2026)
- [Imagination Helps Visual Reasoning, But Not Yet in Latent Space](https://mlsanity.com/paper/2602.22766/) (2026)
- [HyLaR: Hybrid Latent Reasoning with Decoupled Policy Optimization](https://mlsanity.com/paper/2604.20328/) (2026)
- [Visual Latents Know More Than They Say: Unsilencing Latent Reasoning in MLLMs](https://mlsanity.com/paper/2605.02735/) (2026)
- [TVI-CoT: Text-Visual Interleaved Chain-of-Thought Reasoning for Multimodal Understanding](https://mlsanity.com/paper/2606.08464/) (2026)
- [Generative Visual Chain-of-Thought for Image Editing](https://mlsanity.com/paper/2603.01893/) (2026)
- [Gen-VCoT: Generative Visual Chain-of-Thought Reasoning via Diffusion-Based RGB Intermediate Representations](https://mlsanity.com/paper/2606.16783/) (2026)

## Abstract (arXiv)

Interleaved multimodal Chain-of-Thought (CoT) improves visual reasoning by incorporating auxiliary visual evidence into intermediate reasoning. However, existing approaches remain constrained by externally defined reasoning traces and visual operations, limiting their ability to develop flexible and abstract visual thinking. Reasoning with latent has recently offered a promising direction by internalizing intermediate computation into continuous representations. Nevertheless, existing visual-latent methods mainly supervise latent states through alignment with compressed auxiliary visual features, treating them as proxies for visual observations rather than active reasoning states. Consequently, they capture the provided evidence but fail to fully internalize the abstract reasoning process induced by multimodal CoT. In this paper, we propose OPLD (On-Policy Latent Distillation), a simple framework that transfers the reasoning capability induced by privileged multimodal CoT into latent reasoning representations. Extensive experiments on diverse multimodal benchmarks demonstrate that OPLD consistently outperforms existing latent reasoning methods and achieves state-of-the-art performance on multiple benchmarks. The results suggest that supervising latent representations at the reasoning-process level provides a more effective paradigm for multimodal latent reasoning than conventional feature-level alignment.