mlsanityData to 7 Oct 2026

Multimodal reasoning

2,683 papers9 this weekMedium confidence in this grouping: mostly one topic, with blurry edgesMedium confidence in this grouping: mostly one topic, with blurry edges · About

Keywords: visual, multimodal, reasoning, video, visual reasoning

Key facts. Multimodal reasoning: 2,683 papers on arXiv since 2022 (data through 7 Oct 2026). 470 of its 2,683 mapped papers (18%) cite “MathVista: Evaluating Mathematical Reasoning of Foundation Models in Visual Contexts” (2023, arXiv 2310.02255). 300 of its 2,683 mapped papers (11%) cite “Learn to Explain: Multimodal Reasoning via Thought Chains for Science Question Answering” (2022, arXiv 2209.09513). 84 new papers 17 Sep–7 Oct.

What’s new

17 Sep – 7 Oct 2026

84 new papers, 17 Sep – 7 Oct; 9 this week.

  1. This week · 9
  2. Less from More: Reinforcing Sparse Video Reasoning from Dense References
  3. SPLIT-RL: Staged Perception-Language Reasoning Training with Claim-Level Advantages also fits: Reinforcement learning for reasoning
  4. VICO: Visual Environments Co-Evolving for Vision-Language Model Reasoning also fits: Spatial and embodied reasoning
  5. Mixture of Layers: Dynamic Layer Routing for Visual Reasoning
  6. Humanity's Sixth Sense: Benchmarking Intuitive Visual Reasoning in Multimodal Models
  7. Selective Transfer of RL Updates for Visual Reasoning also fits: Reinforcement learning for reasoning
  8. How Well Do LLMs Reason with Noisy Evidence? An Active Visual Reasoning Benchmark also fits: Reasoning benchmarks and evaluation
  9. Matching Object or Relation? Tracing Abstract Reasoning Inside VLMs also fits: Spatial and embodied reasoning
  10. From Transformation to Target State: Rethinking Query Representation for Zero-Shot Composed Image Retrieval
  11. Earlier · 75
  12. When to Rethink: Learning Multi-Perspective Self-Verification for Vision-Language Models
All 84
  1. Atomic Visual Entailment: Enhancing Zero-Shot Vision-Language Reasoning through Atomic Fact Decomposition and Learned Selection
  2. DREAM: Dynamic Resolution Assignment For Multimodal Multi-agent Debate also fits: Reasoning for agents, planning and code
  3. ReMAP: Restoring the Perceptual Cycle with Reasoning-Time Latent Visual Memory
  4. Look Where You Say You're Looking: Self-Grounded Attention for Visual Reasoning
  5. BoT-Feedback: Grounding Multimodal Reasoning in Biomechanical Evidence for Explainable Human Action Feedback also fits: Spatial and embodied reasoning
  6. EgoExo-Next:Benchmarking Vision-Language Models on Visual-Option Next-State and Cross-View Reasoning also fits: Spatial and embodied reasoning
  7. Trinity: Self-Evolving Vision-Language Models with a Self-Verifier
  8. Rethinking Long-Video Efficiency: A Joint Allocation Perspective on Frames, Pixels, and Front-End Latency
  9. CLIMB: Confidence-Guided Complementary Evidence for Multimodal Retrieval-Augmented Generation
  10. Text-Centric Post-Training for Omni-Modal Reasoning also fits: Audio-language model reasoning
  11. SLVR: Structured Latent Visual Reasoning via Human-like Reasoning Flows
  12. VISTA: A Visual Harness for Reasoning in an Interactive World also fits: Spatial and embodied reasoning
  13. MoLE: Mixture of Latent Experts for Complementary Visual Reasoning
  14. Cog-VADU: A Training-Free Cognitive Reasoning Framework for Video Anomaly Detection and Understanding
  15. Hob-VL: A Benchmark for Visually Grounded Boolean Reasoning
  16. Before It Fades: Reinforcing Temporal Representations at Inference Time in VideoLLMs
  17. PRISM: A Category-Theoretic Framework for Measuring and Refining Multimodal Analogies
  18. MMVistaReason: Toward Open-Data and Post-Training Recipes for Multimodal Reasoning
  19. HAWK: Rethinking Multimodal Drafting for Speculative Decoding
  20. Rethinking Multi-Image Re-Representation in Multi-Image Understanding
  21. Aligning Thoughts with Answers: Probability Rewards to Tame Thinking Drift
  22. Reinforcing Multimodal Reasoning via Token-Level Perception-Grounded Advantage Estimation
  23. When Can Text Replace Vision? Structural Bottlenecks in Diagram Reasoning
  24. VideoLoop: Looped Working Memory Against Semantic Thrashing in Long-Form Video Agents
  25. Revisit to Segment: Working Memory Distillation for Reasoning Segmentation
  26. LEGO-OPD: Factorized Teacher Composition for Multimodal On-Policy Distillation
  27. ThinkV2V: Unleashing the Reasoning Capability of MLLMs for Instruction-Guided Video Editing
  28. PhyProbe: Rethinking Physical Consistency Evaluation in Generated Videos
  29. NeuronEye: Query-Guided Visual Concept Activation for Vision-Language Reasoning
  30. SYNCR: Diagnosing and Learning Cross-Video Reasoning from Simulation
  31. Tracing the Evidence: Faithful Token Attribution Through Vision-Language Reasoning
  32. MG-Thinker: Bi-Axial Self-Reflection for Multi-Image Reasoning Grounding
  33. Learning via Self-Consistency for Diffusion-based Video Reasoning also fits: Diffusion reasoning and test-time scaling
  34. FocusVTC: Efficient and High-Performance Visual Text Compression with Adaptive Resolution
  35. ThinkingGuard: Decoding Implicit Hazards via Step-by-Step Risk Attribution in Multimodal Large Language Models also fits: Reasoning benchmarks and evaluation
  36. PARSEE-VAD: Efficient Training-Free Online Video Anomaly Detection via Proposition-Aware Reasoning and Streaming Evidence Escalation
  37. Medical Image Alignment Assessment as a Test of Generalist Visual Reasoning in Frontier Multimodal Models also fits: Clinical and biomedical applications
  38. Question-Specific Knowledge Graphs for Efficient Visual Reasoning
  39. FlowTool: Controlling Tool Parameter in Image Retouching via Flow Matching
  40. Sprout: Building Dynamic Memory While Reasoning for Agentic Video Understanding
  41. Rethinking Visual Token Compression for Video Large Language Models: A Simple Yet Strong Baseline
  42. Token-Disentangled Latent Test-Time Scaling for Vision-Language Reasoning
  43. VD-DeepStack: Bridging Visual Comparison and Language Reasoning for Few-Shot Anomaly Detection
  44. ReSight-SMC: Two-Stage Power Sampling via Island SMC with Visual Scouts
  45. From Perception to Integration: Revisiting the Internal Dynamics of Reasoning in Vision-Language Models
  46. Rethinking Latent Visual Reasoning: Grounding Latent Reasoning in Visual Evidence
  47. Distilling Visual Reasoning into Text Space
  48. SCOPD: Sparse-Context On-Policy Self-Distillation for Efficient Vision-Language Models
  49. MoR-MLLM: Mixture of Recursions for Efficient Multimodal Large Language Models
  50. IVT-Guard: All-in-One Reasoning Model for AI-Generated Content Detection
  51. OPERA: A Unified Omnimodal Progressive Spatio-Temporal Reasoning Agent for Referring Video Segmentation
  52. MM-OPD: Towards One More Bottleneck Between Perception and Reasoning
  53. InterTab: Interleaved Visual-Structure Alignment for Multi-Modal Table Reasoning
  54. SynDORBench: Evaluating LVLM Perceptual Robustness Under Physically Constrained Visibility Conditions
  55. Diagnosing the Sources of Compositional Failure in Vision-Language Models: A Controlled Analysis
  56. Can Linguistic Reasoning Vectors Enhance Multimodal Reasoning Ability?
  57. CCRV-Bench: Constraint-Based Evaluation of Causal Reasoning in Vision-Language Models
  58. MVVBench: Benchmarking 4D Reasoning in Vision-Language Models
  59. Skip the Talk, Re-Focus on Vision: Latent Reasoning for Reasoning Segmentation in Multimodal Large Language Models
  60. Multimodal Thinking with Renderable Programs
  61. Mind What Matters for Reasoning: Aligning Cross-Modal Attention via Selective Probability Mass Concentration
  62. SWE-PolyVision: Benchmarking Cross-Image Abductive Reasoning for Repository-Level Software Engineering
  63. Industrial Anomaly Detection via Defect-Grounded Reasoning in Visual Latent Space
  64. MoVISA: Multi-Token Reasoning for Video Object Segmentation
  65. Looks the Same, Answers Differently: Flip-Direction Steering for Robust Vision-Language Reasoning also fits: Spatial and embodied reasoning
  66. CinematicVQA: Benchmarking Film-Grammar Reasoning in Large Vision-Language Models
  67. From Token Importance to Conditional Removability: Rethinking Visual Token Pruning in Multimodal Large Language Models
  68. Video-HopChain: Multi-Hop Questions and Confidence-Gated Exploration for Video Reasoning Models
  69. PrismGPT: Proxy-Guided Learning for Region-Aware Photo Editing with Self-Synthesized Reasoning
  70. VLM-in-Sandbox: Visual Workspaces for Agentic Visual Reasoning
  71. Look Where It Counts: A Free, Label-Free Visual Evidence Signal for Fine-Grained Vision-Language Reasoning
  72. DRT: Dense Reasoning Trace for Efficient and Grounded Multimodal Reasoning
  73. Beyond Exact Match: Task-Aware GRPO for Cross-Domain PCBA Visual Question Answering
  74. Uni-LaDiR: Latent Diffusion Unifies Multimodal Reasoning

Foundations

Most cited by papers here

  1. MathVista: Evaluating Mathematical Reasoning of Foundation Models in Visual Contexts2023cited by 470 here
  2. Learn to Explain: Multimodal Reasoning via Thought Chains for Science Question Answering2022cited by 300 here
  3. Multimodal Chain-of-Thought Reasoning in Language Models2023cited by 281 here
  4. ChartQA: A Benchmark for Question Answering about Charts with Visual and Logical Reasoning2022cited by 279 here
  5. Measuring Multimodal Mathematical Reasoning with MATH-Vision Dataset2024cited by 277 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 699 here
  2. 2022Self-Consistency Improves Chain of Thought Reasoning in Language Modelscited by 213 here
  3. 2022Large Language Models are Zero-Shot Reasonerscited by 211 here
  4. 2023Multimodal Chain-of-Thought Reasoning in Language Modelscited by 281 here
  5. 2024Visual CoT: Advancing Multi-Modal Language Models with a Comprehensive Dataset and Benchmark for Chain-of-Thought Reasoningcited by 187 here
  6. 2024LLaVA-CoT: Let Vision Language Models Reason Step-by-Stepcited by 236 here
  7. 2025Vision-R1: Incentivizing Reasoning Capability in Multimodal Large Language Modelscited by 260 here
  8. 2025VLM-R1: A Stable and Generalizable R1-style Large Vision-Language Modelcited by 185 here
  9. 2025DeepEyes: Incentivizing "Thinking with Images" via Reinforcement Learningcited by 204 here
15 more it builds on, not in this chain
  1. 2017Proximal Policy Optimization Algorithmscited by 202 here
  2. 2022ChartQA: A Benchmark for Question Answering about Charts with Visual and Logical Reasoningcited by 279 here
  3. 2022Learn to Explain: Multimodal Reasoning via Thought Chains for Science Question Answeringcited by 300 here
  4. 2023LLaMA: Open and Efficient Foundation Language Modelscited by 192 here
  5. 2023Direct Preference Optimization: Your Language Model is Secretly a Reward Modelcited by 165 here
  6. 2023MMBench: Is Your Multi-modal Model an All-around Player?cited by 214 here
  7. 2023MathVista: Evaluating Mathematical Reasoning of Foundation Models in Visual Contextscited by 470 here
  8. 2024DeepSeekMath: Pushing the Limits of Mathematical Reasoning in Open Language Modelscited by 501 here
  9. 2024Measuring Multimodal Mathematical Reasoning with MATH-Vision Datasetcited by 277 here
  10. 2024Expanding Performance Boundaries of Open-Source Multimodal Models with Model, Data, and Test-Time Scalingcited by 230 here
  11. 2025DeepSeek-R1: Incentivizing Reasoning Capability in LLMs via Reinforcement Learningcited by 558 here
  12. 2025MM-Eureka: Exploring the Frontiers of Multimodal Reasoning with Rule-based Reinforcement Learningcited by 169 here
  13. 2025R1-Onevision: Advancing Generalized Multimodal Reasoning through Cross-Modal Formalizationcited by 195 here
  14. 2025Qwen3 Technical Reportcited by 192 here
  15. 2025InternVL3.5: Advancing Open-Source Multimodal Models in Versatility, Reasoning, and Efficiencycited by 175 here

The 24 most cited of 3,271.

Browse

30 of 2,683, ranked 6–35

  1. 6Vision-R1: Incentivizing Reasoning Capability in Multimodal Large Language Models2025260
  2. 7LLaVA-CoT: Let Vision Language Models Reason Step-by-Step2024236
  3. 8Expanding Performance Boundaries of Open-Source Multimodal Models with Model, Data, and Test-Time Scaling2024230
  4. 9MMBench: Is Your Multi-modal Model an All-around Player?2023214
  5. 10DeepEyes: Incentivizing "Thinking with Images" via Reinforcement Learning2025204
  6. 11R1-Onevision: Advancing Generalized Multimodal Reasoning through Cross-Modal Formalization2025195
  7. 12Visual CoT: Advancing Multi-Modal Language Models with a Comprehensive Dataset and Benchmark for Chain-of-Thought Reasoning2024187
  8. 13VLM-R1: A Stable and Generalizable R1-style Large Vision-Language Model2025185
  9. 14InternVL3.5: Advancing Open-Source Multimodal Models in Versatility, Reasoning, and Efficiency2025175
  10. 15MM-Eureka: Exploring the Frontiers of Multimodal Reasoning with Rule-based Reinforcement Learning2025169
All 30
  1. 16Visual-RFT: Visual Reinforcement Fine-Tuning2025158
  2. 17Visual Sketchpad: Sketching as a Visual Chain of Thought for Multimodal Language Models2024157
  3. 18VL-Rethinker: Incentivizing Self-Reflection of Vision-Language Models with Reinforcement Learning2025151
  4. 19Video-R1: Reinforcing Video Reasoning in MLLMs2025139
  5. 20ViperGPT: Visual Inference via Python Execution for Reasoning2023139
  6. 21We-Math: Does Your Large Multimodal Model Achieve Human-like Mathematical Reasoning?2024138
  7. 22LISA: Reasoning Segmentation via Large Language Model2023131
  8. 23Visual Programming: Compositional visual reasoning without training2022130
  9. 24Pixel Reasoner: Incentivizing Pixel-Space Reasoning with Curiosity-Driven Reinforcement Learning2025112
  10. 25Compositional Chain-of-Thought Prompting for Large Multimodal Models2023105
  11. 26LMM-R1: Empowering 3B LMMs with Strong Reasoning Abilities Through Two-Stage Rule-Based RL202597
  12. 27LogicVista: Multimodal LLM Logical Reasoning Benchmark in Visual Contexts202496
  13. 28MM-REACT: Prompting ChatGPT for Multimodal Reasoning and Action202390
  14. 29Imagine while Reasoning in Space: Multimodal Visualization-of-Thought202588
  15. 30Multimodal Chain-of-Thought Reasoning: A Comprehensive Survey202587
  16. 31Kimi-VL Technical Report202585
  17. 32DDCoT: Duty-Distinct Chain-of-Thought Prompting for Multimodal Reasoning in Language Models202384
  18. 33LlamaV-o1: Rethinking Step-by-step Visual Reasoning in LLMs202583
  19. 34OpenVLThinker: Complex Vision-Language Reasoning via Iterative SFT-RL Cycles202580
  20. 35OpenThinkIMG: Learning to Think with Images via Visual Tool Reinforcement Learning202580