mlsanityData to 7 Oct 2026

Diffusion reasoning and test-time scaling

129 papers1 this weekHigh confidence in this grouping: the papers form one clear topicHigh confidence in this grouping: the papers form one clear topic · About

Keywords: diffusion, diffusion language, scaling, inference-time scaling, inference-time

Key facts. Diffusion reasoning and test-time scaling: 129 papers on arXiv since 2022 (data through 7 Oct 2026). 31 of its 129 mapped papers (24%) cite “d1: Scaling Reasoning in Diffusion Large Language Models via Reinforcement Learning” (2025, arXiv 2504.12216). 24 of its 129 mapped papers (19%) cite “A General Framework for Inference-time Scaling and Steering of Diffusion Models” (2025, arXiv 2501.06848). 14 new papers 17 Sep–7 Oct.

What’s new

17 Sep – 7 Oct 2026

14 new papers, 17 Sep – 7 Oct; 1 this week.

  1. This week · 1
  2. Think Before You Paint: Recursive Latent Reasoning for Diffusion Models
  3. Earlier · 13
  4. Preserving Mathematical Reasoning in Compressed Diffusion Language Models via Trajectory-Aware Low-Rank Approximation
  5. Refinement Buys Intelligibility, Search Buys Identity: What Test-Time Compute Buys in Masked-Diffusion TTS
  6. Architectural Sampling: Test-Time Scaling via Computational Diversity in Frozen Vision-Language Models
  7. Exploring More, Reasoning Better: Stepwise Risk-Sensitive GRPO for Diffusion Language Models also fits: Reinforcement learning for reasoning
  8. ExploreNet: Learning Where to Explore in Diffusion GRPO
  9. Rethinking Soft Tokens for Parallel Decoding in Diffusion Language Models
  10. Why Deterministic PRM Guidance Underperforms in Discrete Diffusion Reasoning
  11. ForkLeft: Entropy-First Rollouts for Prefix-Aligned Autoregressive-to-Diffusion Distillation
  12. ELF-REG: Scaling Continuous Diffusion Language Models to Reasoning Tasks
All 14
  1. Towards Efficient Reasoning: Learning Causal Shortcuts for Diffusion Language Models also fits: Causal reasoning in LLMs
  2. Leveraging Inference-Time Compute for Diffusion Models via Global Scheduling of Denoising Trajectories
  3. λ-Controlled GRPO: Turning Flow-Matching Ratio Instability into a Budgeted Resource also fits: Reinforcement learning for reasoning
  4. CaLR: Causal Latent Revision for Robust Diffusion Reasoning

Foundations

Most cited by papers here

  1. d1: Scaling Reasoning in Diffusion Large Language Models via Reinforcement Learning2025cited by 31 here
  2. A General Framework for Inference-time Scaling and Steering of Diffusion Models2025cited by 24 here
  3. Inference-Time Scaling for Diffusion Models beyond Scaling Denoising Steps2025cited by 23 here
  4. wd1: Weighted Policy Optimization for Reasoning in Diffusion Language Models2025cited by 17 here
  5. Remasking Discrete Diffusion Models with Inference-Time Scaling2025cited by 16 here

Lineage

The papers this direction builds on, oldest first.

Each paper on the line cites the one before.

  1. 2022Chain-of-Thought Prompting Elicits Reasoning in Large Language Modelscited by 37 here
  2. 2022Self-Consistency Improves Chain of Thought Reasoning in Language Modelscited by 21 here
  3. 2022Large Language Models are Zero-Shot Reasonerscited by 8 here
  4. 2023Let's Verify Step by Stepcited by 26 here
  5. 2023Math-Shepherd: Verify and Reinforce LLMs Step-by-step without Human Annotationscited by 10 here
  6. 2024DeepSeekMath: Pushing the Limits of Mathematical Reasoning in Open Language Modelscited by 22 here
  7. 2025d1: Scaling Reasoning in Diffusion Large Language Models via Reinforcement Learningcited by 31 here
  8. 2025Reinforcing the Diffusion Chain of Lateral Thought with Diffusion Language Modelscited by 12 here
  9. 2025wd1: Weighted Policy Optimization for Reasoning in Diffusion Language Modelscited by 17 here
15 more it builds on, not in this chain
  1. 2017Proximal Policy Optimization Algorithmscited by 15 here
  2. 2023Tree of Thoughts: Deliberate Problem Solving with Large Language Modelscited by 19 here
  3. 2023Direct Preference Optimization: Your Language Model is Secretly a Reward Modelcited by 17 here
  4. 2024Scaling LLM Test-Time Compute Optimally can be More Effective than Scaling Model Parameterscited by 22 here
  5. 2024Beyond Autoregression: Discrete Diffusion for Complex Reasoning and Planningcited by 15 here
  6. 2024OpenAI o1 System Cardcited by 10 here
  7. 2025A General Framework for Inference-time Scaling and Steering of Diffusion Modelscited by 24 here
  8. 2025Inference-Time Scaling for Diffusion Models beyond Scaling Denoising Stepscited by 23 here
  9. 2025DeepSeek-R1: Incentivizing Reasoning Capability in LLMs via Reinforcement Learningcited by 25 here
  10. 2025SANA 1.5: Efficient Scaling of Training-Time and Inference-Time Compute in Linear Diffusion Transformercited by 10 here
  11. 2025s1: Simple test-time scalingcited by 12 here
  12. 2025Remasking Discrete Diffusion Models with Inference-Time Scalingcited by 16 here
  13. 2025Inference-Time Scaling for Flow Models via Stochastic Generation and Rollover Budget Forcingcited by 10 here
  14. 2025Qwen3 Technical Reportcited by 9 here
  15. 2025Mercury: Ultra-Fast Language Models Based on Diffusioncited by 16 here

The 24 most cited of 344.

Browse

30 of 129, ranked 6–35

  1. 6Mercury: Ultra-Fast Language Models Based on Diffusion202516
  2. 7Beyond Autoregression: Discrete Diffusion for Complex Reasoning and Planning202415
  3. 8Reinforcing the Diffusion Chain of Lateral Thought with Diffusion Language Models202512
  4. 9SANA 1.5: Efficient Scaling of Training-Time and Inference-Time Compute in Linear Diffusion Transformer202510
  5. 10Inference-Time Scaling for Flow Models via Stochastic Generation and Rollover Budget Forcing202510
  6. 11Inference-time Scaling of Diffusion Models through Classical Search20258
  7. 12Diffusion of Thoughts: Chain-of-Thought Reasoning in Diffusion Language Models20247
  8. 13Monte Carlo Tree Diffusion for System 2 Planning20256
  9. 14Improving Reasoning for Diffusion Language Models via Group Diffusion Policy Optimization20256
  10. 15Scaling Image and Video Generation via Test-Time Evolutionary Search20255
All 30
  1. 16The Flexibility Trap: Rethinking the Value of Arbitrary Order in Diffusion Language Models20264
  2. 17Reflect-DiT: Inference-Time Scaling for Text-to-Image Diffusion Transformers via In-Context Reflection20254
  3. 18Adaptive Inference-Time Scaling via Cyclic Diffusion Search20254
  4. 19d2: Improving Reasoning in Diffusion Language Models via Trajectory Likelihood Estimation20254
  5. 20Test-Time Scaling in Diffusion LLMs via Hidden Semi-Autoregressive Experts20254
  6. 21Video-T1: Test-Time Scaling for Video Generation20253
  7. 22TTS-VAR: A Test-Time Scaling Framework for Visual Auto-Regressive Generation20253
  8. 23RFG: Test-Time Scaling for Diffusion Large Language Model Reasoning with Reward-Free Guidance20253
  9. 24Advancing Reasoning in Diffusion Language Models with Denoising Process Rewards20253
  10. 25DiFFPO: Training Diffusion LLMs to Reason Fast and Furious via Reinforcement Learning20253
  11. 26Think While You Generate: Discrete Diffusion with Planned Denoising20243
  12. 27Reasoning with Latent Tokens in Diffusion Language Models20262
  13. 28MetaState: Persistent Working Memory Enhances Reasoning in Discrete Diffusion Language Models20262
  14. 29VFScale: Intrinsic Reasoning through Verifier-Free Test-time Scalable Diffusion Model20252
  15. 30Inference-Time Scaling of Diffusion Language Models via Trajectory Refinement20252
  16. 31DriftLite: Lightweight Drift Control for Inference-Time Scaling of Diffusion Models20252
  17. 32Coevolutionary Continuous Discrete Diffusion: Make Your Diffusion Language Model a Latent Reasoner20252
  18. 33Simple Policy Gradients for Reasoning with Diffusion Language Models20252
  19. 34Enhancing Reasoning for Diffusion LLMs via Distribution Matching Policy Optimization20252
  20. 35Beyond Surface Reasoning: Unveiling the True Long Chain-of-Thought Capacity of Diffusion Large Language Models20252