mlsanityData to 14 Oct 2026

Open-o3-Video: Grounded Video Reasoning with Explicit Spatio-Temporal Evidence

arXiv 2510.20579 ↗First posted 23 Oct 2025cs.CV cs.AI cs.MMcited by 49

On the map: LLM reasoning → Multimodal reasoning

Abstract (arXiv)

Most video reasoning models only generate textual reasoning traces without indicating when and where key evidence appears. Recent models such as OpenAI-o3 have sparked wide interest in evidence-centered reasoning for images, yet extending this ability to videos is more challenging due to the need for joint temporal tracking and spatial localization across dynamic scenes. We introduce Open-o3-Video, a non-agent framework that integrates explicit spatio-temporal evidence into video reasoning by highlighting key timestamps, objects, and bounding boxes, making the reasoning process traceable and verifiable. To enable this capability, we first construct high-quality datasets STGR that provide unified spatio-temporal supervision, which is absent in existing resources. We further adopt a cold-start reinforcement learning strategy with specially designed rewards that jointly encourage answer accuracy, temporal alignment, and spatial precision. On the V-STAR benchmark, Open-o3-Video achieves state-of-the-art performance, improving mAM by 14.4% and mLGM by 24.2% over the Qwen2.5-VL baseline, and shows consistent gains across a range of video understanding benchmarks. Beyond accuracy, the grounded reasoning traces produced by Open-o3-Video support confidence-aware test-time scaling, improving answer reliability.

Key facts. Cited by 49 papers in the map (data through 14 Oct 2026). Placed in Multimodal reasoning, one of 17 directions in LLM reasoning. Builds on 55 of its 66 references that are in the map. First posted on arXiv 23 Oct 2025.

Builds on

55 of 66 references.

Closest ancestors · 19

Same direction, newest first

  1. 2025VTimeCoT: Thinking by Drawing for Video Temporal Grounding and Reasoning16 Oct
  2. 2025Video-MTR: Reinforced Multi-Turn Reasoning for Long Video Understanding28 Aug
  3. 2025Thinking With Videos: Multimodal Tool-Augmented Reinforcement Learning for Long Video Reasoning6 Aug
  4. 2025Scaling RL to Long Videos10 Jul
  5. 2025Video-RTS: Rethinking Reinforcement Learning and Test-Time Scaling for Efficient and Enhanced Video Reasoning9 Jul
  6. 2025VGR: Visual Grounded Reasoning13 Jun
  7. 2025CyberV: Cybernetics for Test-time Scaling in Video Understanding9 Jun
  8. 2025Reinforcing Video Reasoning with Focused Thinking30 May
  9. 2025GRIT: Teaching MLLMs to Think with Images21 May
  10. 2025DeepEyes: Incentivizing "Thinking with Images" via Reinforcement Learning20 May
  11. 2025VideoRFT: Incentivizing Video Reasoning Capability in MLLMs via Reinforced Fine-Tuning18 May
  12. 2025TinyLLaVA-Video-R1: Towards Smaller LMMs for Video Reasoning13 Apr
7 more
  1. 2025VideoChat-R1: Enhancing Spatio-Temporal Perception via Reinforcement Fine-Tuning9 Apr
  2. 2025Video-R1: Reinforcing Video Reasoning in MLLMs27 Mar
  3. 2025V-STaR: Benchmarking Video-LLMs on Video Spatio-Temporal Reasoning14 Mar
  4. 2024Expanding Performance Boundaries of Open-Source Multimodal Models with Model, Data, and Test-Time Scaling6 Dec
  5. 2024VideoEspresso: A Large-Scale Chain-of-Thought Dataset for Fine-Grained Video Reasoning via Core Frame Selection22 Nov
  6. 2024Visual CoT: Advancing Multi-Modal Language Models with a Comprehensive Dataset and Benchmark for Chain-of-Thought Reasoning25 Mar
  7. 2023Look, Remember and Reason: Grounded reasoning in videos with language models30 Jun

Other references · 36

  1. 2024DeepSeekMath: Pushing the Limits of Mathematical Reasoning in Open Language Modelscited by 9,379
  2. 2025Qwen2.5-VL Technical Reportcited by 5,587
  3. 2025Gemini 2.5: Pushing the Frontier with Advanced Reasoning, Multimodality, Long Context, and Next Generation Agentic Capabilitiescited by 3,796
  4. 2024Gemini 1.5: Unlocking multimodal understanding across millions of tokens of contextcited by 3,224
  5. 2025InternVL3: Exploring Advanced Training and Test-Time Recipes for Open-Source Multimodal Modelscited by 1,834
  6. 2023Video-LLaMA: An Instruction-tuned Audio-Visual Language Model for Video Understandingcited by 1,471
  7. 2024Video-MME: The First-Ever Comprehensive Evaluation Benchmark of Multi-modal LLMs in Video Analysiscited by 1,413
  8. 2023MVBench: A Comprehensive Multi-modal Video Understanding Benchmarkcited by 1,109
  9. 2025Group Sequence Policy Optimizationcited by 835
  10. 2017Localizing Moments in Video with Natural Languagecited by 663
  11. 2025VideoLLaMA 3: Frontier Multimodal Foundation Models for Image and Video Understandingcited by 518
  12. 2024Long Context Transfer from Language to Visioncited by 515
24 more
  1. 2019COIN: A Large-scale Dataset for Comprehensive Instructional Video Analysiscited by 318
  2. 2024LongVILA: Scaling Long-Context Visual Language Models for Long Videoscited by 309
  3. 2024mPLUG-Owl3: Towards Long Image-Sequence Understanding in Multi-Modal Large Language Modelscited by 291
  4. 2025Video-MMMU: Evaluating Knowledge Acquisition from Multi-Discipline Professional Videoscited by 236
  5. 2024STAR: A Benchmark for Situated Reasoning in Real-World Videoscited by 213
  6. 2025Adaptive Keyframe Sampling for Long Video Understandingcited by 167
  7. 2024Oryx MLLM: On-Demand Spatial-Temporal Understanding at Arbitrary Resolutioncited by 159
  8. 2025Sa2VA: Marrying SAM2 with MLLM for Dense Grounded Understanding of Images and Videoscited by 159
  9. 2025SpaceR: Reinforcing MLLMs in Video Spatial Reasoningcited by 159
  10. 2025WorldSense: Evaluating Real-world Omnimodal Understanding for Multimodal LLMscited by 155
  11. 2025Time-R1: Post-Training Large Vision Language Model for Temporal Video Groundingcited by 146
  12. 2025PerceptionLM: Open-Access Data and Models for Detailed Visual Understandingcited by 92
  13. 2024TRACE: Temporal Grounding Video LLM via Causal Event Modelingcited by 89
  14. 2025Traceable Evidence Enhanced Visual Grounded Reasoning: Evaluation and Methodologycited by 69
  15. 2025Towards Understanding Camera Motions in Any Videocited by 64
  16. 2025Ross3D: Reconstructive Visual Instruction Tuning with 3D-Awarenesscited by 63
  17. 2025DeepVideo-R1: Video Reinforcement Fine-Tuning via Difficulty-aware Regressive GRPOcited by 61
  18. 2025LLaVA-ST: A Multimodal Large Language Model for Fine-Grained Spatial-Temporal Understandingcited by 59
  19. 2024Reconstructive Visual Instruction Tuningcited by 55
  20. 2022Embracing Consistency: A One-Stage Approach for Spatio-Temporal Video Groundingcited by 42
  21. 2025VideoITG: Multimodal Video Understanding with Instructed Temporal Groundingcited by 29
  22. 2025Universal Video Temporal Grounding with Generative Multi-modal Large Language Modelscited by 27
  23. 2025Datasets and Recipes for Video Temporal Grounding via Reinforcement Learningcited by 19
  24. 2025Fine-grained Spatiotemporal Grounding on Egocentric Videoscited by 11

Built on by

Cited by year (2026 to 14 Oct)

Citing papers per year: 2025: 7; 2026: 42.

49 papers.

Sort:

By area:

  1. 2025OneThinker: All-in-one Reasoning Model for Image and Videocited by 56
  2. 2025Are Video Models Ready as Zero-Shot Reasoners? An Empirical Study with the MME-CoF Benchmarkcited by 43
  3. 2025Skyra: AI-Generated Video Detection via Grounded Artifact Reasoningcited by 25
  4. 2026Video-o3: Native Interleaved Clue Seeking for Long Video Multi-Hop Reasoningcited by 24
  5. 2025AdaTooler-V: Adaptive Tool-Use for Images and Videoscited by 21
  6. 2025REVISOR: Beyond Textual Reflection, Towards Multimodal Introspective Reasoning in Long-Form Video Understandingcited by 16
  7. 2026PyVision-RL: Forging Open Agentic Vision Models via RLcited by 16
  8. 2026SAMTok: Representing Any Mask with Two Wordscited by 11
  9. 2026VISD: Enhancing Video Reasoning via Structured Self-Distillationcited by 10
  10. 2026VideoThinker: Building Agentic VideoLLMs with LLM-Guided Tool Reasoningcited by 9
  11. 2026CamReasoner: Reinforcing Camera Movement Understanding via Structured Spatial Reasoningcited by 9
  12. 2026VideoTemp-o3: Harmonizing Temporal Grounding and Video Understanding in Agentic Thinking-with-Videoscited by 7
37 more
  1. 2026VideoZeroBench: Probing the Limits of Video MLLMs with Spatio-Temporal Evidence Verificationcited by 7
  2. 2026EVA: Efficient Reinforcement Learning for End-to-End Video Agentcited by 6
  3. 2025Know-Show: Benchmarking Video-Language Models on Spatio-Temporal Grounded Reasoningcited by 5
  4. 2026VisionCoach: Reinforcing Grounded Video Reasoning via Visual-Perception Promptingcited by 4
  5. 2026Motion-o: Trajectory-Grounded Video Reasoningcited by 4
  6. 2026StreamingClaw Technical Reportcited by 4
  7. 2026Towards One-to-Many Temporal Groundingcited by 4
  8. 2026Watch, Remember, Reason: Human-View Video Understanding with MLLMscited by 4
  9. 2026Native Active Perception as Reasoning for Omni-Modal Understandingcited by 4
  10. 2026Find, Fix, Reason: Context Repair for Video Reasoningcited by 3
  11. 2026Latent Visual Cache for Video Reasoningcited by 3
  12. 2026OpenVLThinkerV2: A Generalist Multimodal Reasoning Model for Multi-domain Visual Taskscited by 2
  13. 2026VideoKR: Towards Knowledge- and Reasoning-Intensive Video Understandingcited by 2
  14. 2026PixelEyes: Decoupling Perception and Reasoning for Pinpoint Visual Evidence Seekingcited by 2
  15. 2026Evidence-Backed Video Question Answeringcited by 2
  16. 2026DynTrace: Tracking Dynamic Object Evidence for 4D Spatio-Temporal Reasoning in MLLMscited by 2
  17. 2026Audio-Visual Flamingo: Open Audio-Visual Intelligence for Long and Complex Videoscited by 2
  18. 2026DynFrame: Adaptive Reasoning-Driven Multimodal Framework with Dynamic Frame Augmentation for Complex Video Understandingcited by 1
  19. 2026PerceptionDLM: Parallel Region Perception with Multimodal Diffusion Language Modelscited by 1
  20. 2026MotionAtlas: Detailed Region Captioning for Motion-Centric Videoscited by 1
  21. 2025Decoupling Perception from Reasoning for Hallucination-Resistant Video Understandingcited by 0
  22. 2026Beyond Single-Sample: Reliable Multi-Sample Distillation for Video Understandingcited by 0
  23. 2026One Identity, Many Roles: Multimodal Entity Coreference for Enhanced Video Situation Recognitioncited by 0
  24. 2026CaST-Bench: Benchmarking Causal Chain-Grounded Spatio-Temporal Reasoning for Video Question Answeringcited by 0
  25. 2026VideoLatent: Video-Language Learning via Latent Self-Forcingcited by 0
  26. 2026SER: Learning to Ground Video Reasoning with Semantic Evidence Rewardscited by 0
  27. 2026Efficient Spatio-Temporal Grounding with Multimodal Large Models via Second-Level Tracking and RL Verificationcited by 0
  28. 2026AnyGroundBench: A Specialized-Domain Benchmark for Video Grounding in Vision-Language Modelscited by 0
  29. 2026Perception Before Reasoning: Dynamic Latent Reasoning for Video Understanding and Question Answeringcited by 0
  30. 2026Video-FLAIR: Not Whether to Reason, But Howcited by 0
  31. 2026VideoScout: Learning Agentic Active Exploration with Adaptive Reasoning Pacing for Long Video Understandingcited by 0
  32. 2026InterTab: Interleaved Visual-Structure Alignment for Multi-Modal Table Reasoningcited by 0
  33. 2026MeSD: Multi-Evidence Self-Distillation for VideoLLMcited by 0
  34. 2026Open-o3-Video-v2: Seek, Ground, and Reason with Verifiable Spatio-Temporal Evidencecited by 0
  35. 2026Meta-Video: Efficient Standardized Evaluation for Video Understandingcited by 0
  36. 2026OmniCompass: A Holistic Benchmark for Omni-Modal Video Grounding and Understandingcited by 0
  37. 2026VITAL: From Visual Motion to Precise Actioncited by 0

Similar, not cited

Close in wording, no citation either way

  1. 2026Reasoning over Video: Evaluating How MLLMs Extract, Integrate, and Reconstruct Spatiotemporal Evidence
  2. 2025Video Spatial Reasoning with Object-Centric 3D Rollout
  3. 2026StateTrace: An Object-Centric Framework for Hidden-State Spatiotemporal Reasoning in Long Videos
  4. 2026VideoSEG-O3: A Multi-turn Reinforcement Learning Framework for Reasoning Video Object Segmentation
  5. 2025Video-STR: Reinforcing MLLMs in Video Spatio-Temporal Reasoning with Relation Graph
  6. 2026VGenST-Bench: A Benchmark for Spatio-Temporal Reasoning via Active Video Synthesis
  7. 2025Video-LMM Post-Training: A Deep Dive into Video Reasoning with Large Multimodal Models
  8. 2026STORM: Internalized Modeling for Spatial-Temporal Reasoning in Video-Language Models