用扩散模型做视觉主导的多模态推理，效果远超传统文本中心模型。
DiffThinker: Towards Generative Multimodal Reasoning with Diffusion Models
- 将多模态推理转为图像到图像生成任务，提升空间精度与逻辑一致性。
- 在4个领域测试中，性能比GPT-5高314.2%，比Gemini-3-Flash高111.6%。
- 适合需要精确视觉推理的场景，如规划、布局和组合优化任务。
尽管近期多模态大语言模型（MLLMs）在多模态推理上取得显著进展，其推理过程仍以文本为中心，导致在复杂长时程、视觉主导任务中表现不佳。本文提出一种新的生成式多模态推理范式，并引入基于扩散模型的DiffThinker框架。概念上，DiffThinker将多模态推理重构为原生的图像到图像生成任务，在视觉主导任务中实现更优的逻辑一致性和空间精度。我们系统对比了DiffThinker与MLLMs，首次深入揭示该范式的四大核心特性：效率、可控性、原生并行性与协作性。在四个领域（序列规划、组合优化、约束满足、空间配置）的大量实验表明，DiffThinker显著优于领先闭源模型，包括GPT-5（+314.2%）、Gemini-3-Flash（+111.6%），以及微调后的Qwen3-VL-32B基线（+39.0%），凸显生成式多模态推理在视觉主导推理中的巨大潜力。
原文摘要 · Abstract (English)
While recent Multimodal Large Language Models (MLLMs) have attained significant strides in multimodal reasoning, their reasoning processes remain predominantly text-centric, leading to suboptimal performance in complex long-horizon, vision-centric tasks. In this paper, we establish a novel Generative Multimodal Reasoning paradigm and introduce DiffThinker, a diffusion-based reasoning framework. Conceptually, DiffThinker reformulates multimodal reasoning as a native generative image-to-image task, achieving superior logical consistency and spatial precision in vision-centric tasks. We perform a systematic comparison between DiffThinker and MLLMs, providing the first in-depth investigation into the intrinsic characteristics of this paradigm, revealing four core properties: efficiency, controllability, native parallelism, and collaboration. Extensive experiments across four domains (sequential planning, combinatorial optimization, constraint satisfaction, and spatial configuration) demonstrate that DiffThinker significantly outperforms leading closed source models including GPT-5 (+314.2\%) and Gemini-3-Flash (+111.6\%), as well as the fine-tuned Qwen3-VL-32B baseline (+39.0\%), highlighting generative multimodal reasoning as a promising approach for vision-centric reasoning.
Thank you to arXiv for use of its open access interoperability. PaperDance 不是 arXiv 官方产品；中文卡片由大模型生成，请以原文为准。