mlsanityData to 7 Oct 2026

OneThinker: All-in-one Reasoning Model for Image and Video

arXiv 2512.03043 ↗First posted 2 Dec 2025cs.CVcited by 52

On the map: LLM reasoning → Multimodal reasoning

also in 1 more
Abstract (arXiv)

Reinforcement learning (RL) has recently achieved remarkable success in eliciting visual reasoning within Multimodal Large Language Models (MLLMs). However, existing approaches typically train separate models for different tasks and treat image and video reasoning as disjoint domains. This results in limited scalability toward a multimodal reasoning generalist, which restricts practical versatility and hinders potential knowledge sharing across tasks and modalities. To this end, we propose OneThinker, an all-in-one reasoning model that unifies image and video understanding across diverse fundamental visual tasks, including question answering, captioning, spatial and temporal grounding, tracking, and segmentation. To achieve this, we construct the OneThinker-600k training corpus covering all these tasks and employ commercial models for CoT annotation, resulting in OneThinker-SFT-340k for SFT cold start. Furthermore, we propose EMA-GRPO to handle reward heterogeneity in multi-task RL by tracking task-wise moving averages of reward standard deviations for balanced optimization. Extensive experiments on diverse visual benchmarks show that OneThinker delivers strong performance on 31 benchmarks, across 10 fundamental visual understanding tasks. Moreover, it exhibits effective knowledge transfer between certain tasks and preliminary zero-shot generalization ability, marking a step toward a unified multimodal reasoning generalist. All code, model, and data are released.

Key facts. Cited by 52 papers in the map (data through 7 Oct 2026). Placed in Multimodal reasoning, one of 17 directions in LLM reasoning. Builds on 82 of its 86 references that are in the map. First posted on arXiv 2 Dec 2025.

Builds on

82 of 86 references.

Closest ancestors · 31

Same direction, newest first

  1. 2025Open-o3-Video: Grounded Video Reasoning with Explicit Spatio-Temporal Evidence23 Oct
  2. 2025CodePlot-CoT: Mathematical Visual Reasoning by Thinking with Code-Driven Images13 Oct
  3. 2025More Thought, Less Accuracy? On the Dual Nature of Reasoning in Vision-Language Models30 Sep
  4. 2025Thinking With Videos: Multimodal Tool-Augmented Reinforcement Learning for Long Video Reasoning6 Aug
  5. 2025Scaling RL to Long Videos10 Jul
  6. 2025Seg-R1: Segmentation Can Be Surprisingly Simple with Reinforcement Learning27 Jun
  7. 2025VideoMathQA: Benchmarking Mathematical Reasoning via Multimodal Understanding in Videos5 Jun
  8. 2025Advancing Multimodal Reasoning: From Optimized Cold Start to Staged Reinforcement Learning4 Jun
  9. 2025Reinforcement Learning Tuning for VideoLLMs: Reward Design and Data Efficiency2 Jun
  10. 2025ProxyThinker: Test-Time Guidance through Small Visual Reasoners30 May
  11. 2025Video-Holmes: Can MLLM Think Like Holmes for Complex Video Reasoning?27 May
  12. 2025MME-Reasoning: A Comprehensive Benchmark for Logical Reasoning in MLLMs27 May
19 more
  1. 2025Reinforcement Fine-Tuning Powers Reasoning Capability of Multimodal Large Language Models24 May
  2. 2025DeepEyes: Incentivizing "Thinking with Images" via Reinforcement Learning20 May
  3. 2025Reinforced MLLM: A Survey on RL-Based Reasoning in Multimodal Large Language Models30 Apr
  4. 2025VL-Rethinker: Incentivizing Self-Reflection of Vision-Language Models with Reinforcement Learning10 Apr
  5. 2025Perception-R1: Pioneering Perception Policy with Reinforcement Learning10 Apr
  6. 2025VLM-R1: A Stable and Generalizable R1-style Large Vision-Language Model10 Apr
  7. 2025VideoChat-R1: Enhancing Spatio-Temporal Perception via Reinforcement Fine-Tuning9 Apr
  8. 2025Video-R1: Reinforcing Video Reasoning in MLLMs27 Mar
  9. 2025MM-Eureka: Exploring the Frontiers of Multimodal Reasoning with Rule-based Reinforcement Learning10 Mar
  10. 2025Vision-R1: Incentivizing Reasoning Capability in Multimodal Large Language Models9 Mar
  11. 2025Seg-Zero: Reasoning-Chain Guided Segmentation via Cognitive Reinforcement9 Mar
  12. 2024One Token to Seg Them All: Language Instructed Reasoning Segmentation in Videos29 Sep
  13. 2024VISA: Reasoning Video Object Segmentation via Large Language Models16 Jul
  14. 2023PixelLM: Pixel Reasoning with Large Multimodal Model4 Dec
  15. 2023MathVista: Evaluating Mathematical Reasoning of Foundation Models in Visual Contexts3 Oct
  16. 2023LISA: Reasoning Segmentation via Large Language Model1 Aug
  17. 2023MMBench: Is Your Multi-modal Model an All-around Player?12 Jul
  18. 2023Multimodal Chain-of-Thought Reasoning in Language Models2 Feb
  19. 2022Learn to Explain: Multimodal Reasoning via Thought Chains for Science Question Answering20 Sep

Other references · 51

  1. 2024GPT-4o System Cardcited by 4,467
  2. 2023Improved Baselines with Visual Instruction Tuningcited by 4,362
  3. 2025Gemini 2.5: Pushing the Frontier with Advanced Reasoning, Multimodality, Long Context, and Next Generation Agentic Capabilitiescited by 3,655
  4. 2024SAM 2: Segment Anything in Images and Videoscited by 2,784
  5. 2025DAPO: An Open-Source LLM Reinforcement Learning System at Scalecited by 2,678
  6. 2025Qwen3-VL Technical Reportcited by 2,534
  7. 2023MMMU: A Massive Multi-discipline Multimodal Understanding and Reasoning Benchmark for Expert AGIcited by 2,044
  8. 2024Video-MME: The First-Ever Comprehensive Evaluation Benchmark of Multi-modal LLMs in Video Analysiscited by 1,375
  9. 2025Understanding R1-Zero-Like Training: A Critical Perspectivecited by 1,276
  10. 2016Modeling Context in Referring Expressionscited by 1,054
  11. 2024Are We on the Right Way for Evaluating Large Vision-Language Models?cited by 893
  12. 2016A Diagram Is Worth A Dozen Imagescited by 793
38 more
  1. 2017Dense-Captioning Events in Videoscited by 789
  2. 2025Group Sequence Policy Optimizationcited by 788
  3. 2024LongVideoBench: A Benchmark for Long-context Interleaved Video-Language Understandingcited by 644
  4. 2017TALL: Temporal Activity Localization via Language Querycited by 581
  5. 2024Thinking in Space: How Multimodal Large Language Models See, Remember, and Recall Spacescited by 580
  6. 2024MathVerse: Does Your Multi-modal LLM Truly See the Diagrams in Visual Math Problems?cited by 573
  7. 2025VideoLLaMA 3: Frontier Multimodal Foundation Models for Image and Video Understandingcited by 502
  8. 2018GOT-10k: A Large High-Diversity Benchmark for Generic Object Tracking in the Wildcited by 471
  9. 2025Group-in-Group Policy Optimization for LLM Agent Trainingcited by 442
  10. 2025Towards Reasoning Era: A Survey of Long Chain-of-Thought for Reasoning Large Language Modelscited by 381
  11. 2025Seed1.5-VL Technical Reportcited by 367
  12. 2025Logic-RL: Unleashing LLM Reasoning with Rule-Based Reinforcement Learningcited by 260
  13. 2023VTimeLLM: Empower LLM to Grasp Video Momentscited by 253
  14. 2023MeViS: A Large-scale Benchmark for Video Segmentation with Motion Expressionscited by 225
  15. 2025Video-MMMU: Evaluating Knowledge Acquisition from Multi-Discipline Professional Videoscited by 225
  16. 2025InternVideo2.5: Empowering Video MLLMs with Long and Rich Context Modelingcited by 203
  17. 2024MMT-Bench: A Comprehensive Multimodal Benchmark for Evaluating Large Vision-Language Models Towards Multitask AGIcited by 192
  18. 2022Language as Queries for Referring Video Object Segmentationcited by 180
  19. 2025Agentic Reinforced Policy Optimizationcited by 166
  20. 2025GPG: A Simple and Strong Reinforcement Learning Baseline for Model Reasoningcited by 147
  21. 2025Time-R1: Post-Training Large Vision Language Model for Temporal Video Groundingcited by 138
  22. 2025MMVU: Measuring Expert-Level Multi-Discipline Video Understandingcited by 131
  23. 2025Reinforcing Spatial Reasoning in Vision-Language Models with Interwoven Thinking and Visual Drawingcited by 130
  24. 2024LITA: Language Instructed Temporal-Localization Assistantcited by 124
  25. 2024GroundingGPT:Language Enhanced Multi-modal Grounding Modelcited by 101
  26. 2024TimeSuite: Improving MLLMs for Long Video Understanding via Grounded Tuningcited by 99
  27. 2024Grounded-VideoLLM: Sharpening Fine-grained Temporal Grounding in Video Large Language Modelscited by 87
  28. 2025Critique-GRPO: Advancing LLM Reasoning with Natural Language and Numerical Feedbackcited by 86
  29. 2025LLaVA-ST: A Multimodal Large Language Model for Fine-Grained Spatial-Temporal Understandingcited by 57
  30. 2025SophiaVL-R1: Reinforcing MLLMs Reasoning with Thinking Rewardcited by 45
  31. 2025Video-MMLU: A Massive Multi-Discipline Lecture Understanding Benchmarkcited by 37
  32. 2025STAR-R1: Spatial TrAnsformation Reasoning by Reinforcing Multimodal LLMscited by 21
  33. 2025Affordance-R1: Reinforcement Learning for Generalizable Affordance Reasoning in Multimodal Large Language Modelcited by 21
  34. 2025Uncalibrated Reasoning: GRPO Induces Overconfidence for Stochastic Outcomescited by 21
  35. 2025SpaceVista: All-Scale Visual Spatial Reasoning from mm to kmcited by 18
  36. 2025ARES: Multimodal Adaptive Reasoning via Difficulty-Aware Token-Level Entropy Shapingcited by 16
  37. 2025Pre-Trained Policy Discriminators are General Reward Modelscited by 14
  38. 2025MAPO: Mixed Advantage Policy Optimizationcited by 11

50 most cited of 51 All 51 (JSON)

Built on by

Cited by year (2026 to 7 Oct)

Citing papers per year: 2025: 4; 2026: 48.

52 papers.

Sort:

By area:

  1. 2026Gen-Searcher: Reinforcing Agentic Search for Image Generationcited by 25
  2. 2025EditThinker: Unlocking Iterative Reasoning for Any Image Editorcited by 23
  3. 2025AdaTooler-V: Adaptive Tool-Use for Images and Videoscited by 20
  4. 2026Inference-Time Scaling of Verification: Self-Evolving Deep Research Agents via Test-Time Rubric-Guided Verificationcited by 19
  5. 2026Video Streaming Thinking: VideoLLMs Can Watch and Think Simultaneouslycited by 19
  6. 2026OpenSearch-VL: An Open Recipe for Frontier Multimodal Search Agentscited by 18
  7. 2026Vision-aligned Latent Reasoning for Multi-modal Large Language Modelcited by 16
  8. 2026VideoVeritas: AI-Generated Video Detection via Perception Pretext Reinforcement Learningcited by 11
  9. 2026Wall-OSS-0.5 Technical Reportcited by 11
  10. 2026Vero: An Open RL Recipe for General Visual Reasoningcited by 10
  11. 2026Dual Latent Memory for Visual Multi-agent Systemcited by 9
  12. 2026CamReasoner: Reinforcing Camera Movement Understanding via Structured Spatial Reasoningcited by 8
38 more
  1. 2026Seg-ReSearch: Segmentation with Interleaved Reasoning and External Searchcited by 6
  2. 2026AdaCodec: A Predictive Visual Code for Video MLLMscited by 6
  3. 2025A Benchmark for Omni-Modal Reasoning in Long Videoscited by 5
  4. 2026From Seeing to Thinking: Decoupling Perception and Reasoning Improves Post-Training of Vision-Language Modelscited by 5
  5. 2026Embodied-R1.5: Evolving Physical Intelligence via Embodied Foundation Modelscited by 5
  6. 2026Co-Evolving Policy Distillationcited by 4
  7. 2026Visual Enhanced Depth Scaling for Multimodal Latent Reasoningcited by 3
  8. 2026Persistent Visual Memory: Sustaining Perception for Deep Generation in LVLMscited by 3
  9. 2026From Web to Pixels: Bringing Agentic Search into Visual Perceptioncited by 3
  10. 2026Temporal-Aware Reasoning Optimization for Video Temporal Groundingcited by 3
  11. 2026CoLT: Teaching Multi-Modal Models to Think with Chain of Latent Thoughtscited by 3
  12. 2026OpenVLThinkerV2: A Generalist Multimodal Reasoning Model for Multi-domain Visual Taskscited by 2
  13. 2026Overcoming Catastrophic Forgetting in Visual Continual Learning with Reinforcement Fine-Tuningcited by 2
  14. 2025From Macro to Micro: Benchmarking Microscopic Spatial Intelligence on Molecules via Vision-Language Modelscited by 1
  15. 2026Chatting with Images for Introspective Visual Thinkingcited by 1
  16. 2026Chain-of-Glimpse: Search-Guided Progressive Object-Grounded Reasoning for Video Understandingcited by 1
  17. 2026EasyVideoR1: Easier RL for Video Understandingcited by 1
  18. 2026Agentic AI for Remote Sensing: Technical Challenges and Research Directionscited by 1
  19. 2026Perception Without Engagement: Dissecting the Causal Discovery Deficit in LMMscited by 1
  20. 2026Don't Guess, Just Ask: Resolving Ambiguity in Referring Segmentation via Multi-turn Clarificationcited by 1
  21. 2026Learning Spatiotemporal Sensitivity in Video LLMs via Counterfactual Reinforcement Learningcited by 1
  22. 2026VideoKR: Towards Knowledge- and Reasoning-Intensive Video Understandingcited by 1
  23. 2026UniCoder: Unified Visual-to-Code Generation via Symbolic Rewards and Reference-Guided Code Optimizationcited by 1
  24. 2026VideoSearch-R1: Iterative Video Retrieval and Reasoning via Soft Query Refinementcited by 1
  25. 2026Annotations as Rollouts: Efficient and Scalable Reinforcement Learning for Video MLLMscited by 1
  26. 2026Video-MOPD: Multi-Teacher On-Policy Distillation for Video Understandingcited by 1
  27. 2026OmniSapiens: A Foundation Model for Social Behavior Processing via Heterogeneity-Aware Relative Policy Optimizationcited by 0
  28. 2026Dual Tuning for Reasoning Efficacy-Driven Data Curation in Multimodal LLM Trainingcited by 0
  29. 2026VersusQ: Pairwise Margin Reasoning for Generalizable Video Quality Assessmentcited by 0
  30. 2026MindHelper: Closed-Loop Embodied Mental-State Reasoning for Precision Interventioncited by 0
  31. 2026VideoSEG-O3: A Multi-turn Reinforcement Learning Framework for Reasoning Video Object Segmentationcited by 0
  32. 2026CoRe: A Comprehensive Framework for Cross-Image Comparative Reasoning in Vision-Language Modelscited by 0
  33. 2026Capek 0.5: An Execution-Centric Vision-Language Model for Embodied Intelligencecited by 0
  34. 2026AVCap: Reinforcing Audio-Video Joint Caption with Detail-Aware Rewardcited by 0
  35. 2026Video-FLAIR: Not Whether to Reason, But Howcited by 0
  36. 2026SVMemAgent: A Streaming Video Memory Agent for Query-Agnostic Online Frame Selectioncited by 0
  37. 2026Video-HopChain: Multi-Hop Questions and Confidence-Gated Exploration for Video Reasoning Modelscited by 0
  38. 2026Fewer Tokens, More Self-Teaching: On-Policy Self-Distillation for Extreme Visual Token Reductioncited by 0

50 most cited of 52 All 52 (JSON)

Similar, not cited

Close in wording, no citation either way

  1. 2025R1-Onevision: Advancing Generalized Multimodal Reasoning through Cross-Modal Formalization
  2. 2025ViSS-R1: Self-Supervised Reinforcement Video Reasoning
  3. 2025Video-Thinker: Sparking "Thinking with Videos" via Reinforcement Learning
  4. 2025FrameThinker: Learning to Think with Long Videos via Multi-Turn Frame Spotlighting
  5. 2025Constructing and Interpreting Digital Twin Representations for Visual Reasoning via Reinforcement Learning
  6. 2025Reinforcing Video Reasoning with Focused Thinking
  7. 2026See It, Say It, Sorted: An Iterative Training-Free Framework for Visually-Grounded Multimodal Reasoning in LVLMs
  8. 2025GThinker: Towards General Multimodal Reasoning via Cue-Guided Rethinking