mlsanityData to 7 Oct 2026

Cambrian-S: Towards Spatial Supersensing in Video

arXiv 2511.04670 ↗First posted 6 Nov 2025cs.CVcited by 123

On the map: World models → Video-generation world models and simulators

Abstract (arXiv)

We argue that progress in true multimodal intelligence calls for a shift from reactive, task-driven systems and brute-force long context towards a broader paradigm of supersensing. We frame spatial supersensing as four stages beyond linguistic-only understanding: semantic perception (naming what is seen), streaming event cognition (maintaining memory across continuous experiences), implicit 3D spatial cognition (inferring the world behind pixels), and predictive world modeling (creating internal models that filter and organize information). Current benchmarks largely test only the early stages, offering narrow coverage of spatial cognition and rarely challenging models in ways that require true world modeling. To drive progress in spatial supersensing, we present VSI-SUPER, a two-part benchmark: VSR (long-horizon visual spatial recall) and VSC (continual visual spatial counting). These tasks require arbitrarily long video inputs yet are resistant to brute-force context expansion. We then test data scaling limits by curating VSI-590K and training Cambrian-S, achieving +30% absolute improvement on VSI-Bench without sacrificing general capabilities. Yet performance on VSI-SUPER remains limited, indicating that scale alone is insufficient for spatial supersensing. We propose predictive sensing as a path forward, presenting a proof-of-concept in which a self-supervised next-latent-frame predictor leverages surprise (prediction error) to drive memory and event segmentation. On VSI-SUPER, this approach substantially outperforms leading proprietary baselines, showing that spatial supersensing requires models that not only see but also anticipate, select, and organize experience.

Key facts. Cited by 123 papers in the map (data through 7 Oct 2026). Placed in Video-generation world models and simulators, one of 11 directions in World models. Builds on 39 of its 168 references that are in the map. First posted on arXiv 6 Nov 2025.

Builds on

39 of 168 references.

References · 39

  1. 2023GPT-4 Technical Reportcited by 18,132
  2. 2023LLaMA: Open and Efficient Foundation Language Modelscited by 13,958
  3. 2023Llama 2: Open Foundation and Fine-Tuned Chat Modelscited by 12,865
  4. 2025Qwen2.5-VL Technical Reportcited by 5,418
  5. 2024Qwen2.5 Technical Reportcited by 4,713
  6. 2024GPT-4o System Cardcited by 4,467
  7. 2024Qwen2-VL: Enhancing Vision-Language Model's Perception of the World at Any Resolutioncited by 4,151
  8. 2025Gemini 2.5: Pushing the Frontier with Advanced Reasoning, Multimodality, Long Context, and Next Generation Agentic Capabilitiescited by 3,655
  9. 2016Gaussian Error Linear Units (GELUs)cited by 3,431
  10. 2023Qwen Technical Reportcited by 3,273
  11. 2024Gemini 1.5: Unlocking multimodal understanding across millions of tokens of contextcited by 3,175
  12. 2024OpenVLA: An Open-Source Vision-Language-Action Modelcited by 2,777
27 more
  1. 2023Qwen-VL: A Versatile Vision-Language Model for Understanding, Localization, Text Reading, and Beyondcited by 2,179
  2. 2025InternVL3: Exploring Advanced Training and Test-Time Recipes for Open-Source Multimodal Modelscited by 1,769
  3. 2020Linformer: Self-Attention with Linear Complexitycited by 1,462
  4. 2018World Modelscited by 1,444
  5. 2025SigLIP 2: Multilingual Vision-Language Encoders with Improved Semantic Understanding, Localization, and Dense Featurescited by 1,127
  6. 2023VideoChat: Chat-Centric Video Understandingcited by 851
  7. 2023Gemini: A Family of Highly Capable Multimodal Modelscited by 810
  8. 2025V-JEPA 2: Self-Supervised Video Models Enable Understanding, Prediction and Planningcited by 738
  9. 2024Long Context Transfer from Language to Visioncited by 502
  10. 2025SmolVLM: Redefining small and efficient multimodal modelscited by 226
  11. 2025Video-MMMU: Evaluating Knowledge Acquisition from Multi-Discipline Professional Videoscited by 225
  12. 2024VideoChat-Flash: Hierarchical Compression for Long-Context Video Modelingcited by 212
  13. 2025SpaceR: Reinforcing MLLMs in Video Spatial Reasoningcited by 152
  14. 2024VideoGPT+: Integrating Image and Video Encoders for Enhanced Video Understandingcited by 113
  15. 2021GSPMD: General and Scalable Parallelization for ML Computation Graphscited by 109
  16. 2025MindCube: Spatial Mental Modeling from Limited Viewscited by 91
  17. 2025Gemini Robotics: Bringing AI into the Physical Worldcited by 77
  18. 2025Intuitive physics understanding emerges from self-supervised pretraining on natural videoscited by 66
  19. 2025Multi-SpatialMLLM: Multi-Frame Spatial Understanding with Multi-Modal Large Language Modelscited by 52
  20. 2025MindJourney: Test-Time Scaling with World Models for Spatial Reasoningcited by 49
  21. 2025STORM: Token-Efficient Long Video Understanding for Multimodal LLMscited by 46
  22. 2024Needle In A Video Haystack: A Scalable Synthetic Evaluator for Video MLLMscited by 35
  23. 2025Whole-Body Conditioned Egocentric Video Predictioncited by 30
  24. 2024COSMO: COntrastive Streamlined MultimOdal Model with Interleaved Pre-Trainingcited by 18
  25. 2025Seeing from Another Perspective: Evaluating Multi-View Understanding in MLLMscited by 16
  26. 2025ZoomV: Temporal Zoom-in for Efficient Long Video Understandingcited by 10
  27. 2025NeMo: Needle in a Montage for Video-Language Understandingcited by 3

Built on by

Cited by year (2026 to 7 Oct)

Citing papers per year: 2025: 13; 2026: 110.

123 papers.

Sort:

By area:

  1. 2025Scaling Spatial Intelligence with Multimodal Foundation Modelscited by 75
  2. 2026MolmoAct2: Action Reasoning Models for Real-world Deploymentcited by 45
  3. 2025SpatialBench: Benchmarking Multimodal Large Language Models for Spatial Cognitioncited by 35
  4. 2026Beyond Language Modeling: An Exploration of Multimodal Pretrainingcited by 32
  5. 2026LLaVA-OneVision-2: Towards Next-Generation Perceptual Intelligencecited by 31
  6. 2026InternVL-U: Democratizing Unified Multimodal Models for Understanding, Reasoning, Generation and Editingcited by 23
  7. 2026Think3D: Thinking with Space for Spatial Reasoningcited by 22
  8. 2025QuantiPhy: A Quantitative Benchmark Evaluating Physical Reasoning Abilities of Vision-Language Modelscited by 19
  9. 2026Utonia: Toward One Encoder for All Point Cloudscited by 18
  10. 2026TwinBrainVLA: Unleashing the Potential of Generalist VLMs for Embodied Tasks via Asymmetric Mixture-of-Transformerscited by 14
  11. 2026Thinking with Geometry: Active Geometry Integration for Spatial Reasoningcited by 14
  12. 2026TimeBlind: A Spatio-Temporal Compositionality Benchmark for Video LLMscited by 11
38 more
  1. 2026SpatialStack: Layered Geometry-Language Fusion for 3D VLM Spatial Reasoningcited by 11
  2. 2026ReVSI: Rebuilding Visual Spatial Intelligence Evaluation for Accurate Assessment of VLM 3D Reasoningcited by 11
  3. 2026EgoMemReason: A Memory-Driven Reasoning Benchmark for Long-Horizon Egocentric Video Understandingcited by 11
  4. 2026Lance: Unified Multimodal Modeling by Multi-Task Synergycited by 11
  5. 2025Learning to Reason in 4D: Dynamic Spatial Understanding for Vision Language Modelscited by 10
  6. 2026Imagination Helps Visual Reasoning, But Not Yet in Latent Spacecited by 10
  7. 2025InfiniteVL: Synergizing Linear and Sparse Attention for Highly-Efficient, Unlimited-Input Vision-Language Modelscited by 9
  8. 2025Beyond Flatlands: Unlocking Spatial Intelligence by Decoupling 3D Reasoning from Numerical Regressioncited by 7
  9. 2026Can LLMs See Without Pixels? Benchmarking Spatial Intelligence from Textual Descriptionscited by 7
  10. 2026CausalSpatial: A Benchmark for Object-Centric Causal Spatial Reasoningcited by 7
  11. 2026Done, But Not Sure: Disentangling World Completion from Self-Termination in Embodied Agentscited by 7
  12. 2026S-Agent: Spatial Tool-Use Elicits Reasoning for Spatial Intelligencecited by 7
  13. 2026Spa3R: Predictive Spatial Field Modeling for 3D Visual Reasoningcited by 6
  14. 2026GEM: Generative Supervision Helps Embodied Intelligencecited by 6
  15. 2026From Pixels to Words -- Towards Native One-Vision Models at Scalecited by 6
  16. 2025VideoSSR: Video Self-Supervised Reinforcement Learningcited by 5
  17. 2026Video-MSR: Benchmarking Multi-hop Spatial Reasoning Capabilities of MLLMscited by 5
  18. 2026Holi-Spatial: Evolving Video Streams into Holistic 3D Spatial Intelligencecited by 5
  19. 2026Video Understanding: From Geometry and Semantics to Unified Modelscited by 5
  20. 2026PanoWorld: Towards Spatial Supersensing in 360^circ Panorama Worldcited by 5
  21. 2026Embodied-R1.5: Evolving Physical Intelligence via Embodied Foundation Modelscited by 5
  22. 2025Video2Layout: Recall and Reconstruct Metric-Grounded Cognitive Map for Spatial Reasoningcited by 4
  23. 2026OmniStream: Mastering Perception, Reconstruction and Action in Continuous Streamscited by 4
  24. 2026OpenSpatial: A Principled Data Engine for Empowering Spatial Intelligencecited by 4
  25. 2026Nano World Models: A Minimalist Implementation of Future Video Predictioncited by 4
  26. 2026OVO-S-Bench: A Hierarchical Benchmark for Streaming Spatial Intelligence in Multimodal LLMscited by 4
  27. 2025Scaling Spatial Reasoning in MLLMs through Programmatic Data Synthesiscited by 3
  28. 2025Spatial4D-Bench: A Versatile 4D Spatial Intelligence Benchmarkcited by 3
  29. 2026Research on World Models Is Not Merely Injecting World Knowledge into Specific Taskscited by 3
  30. 2026See, Remember, Explore: A Benchmark and Baselines for Streaming Spatial Reasoningcited by 3
  31. 2026Are VLMs Lost Between Sky and Space? LinkS2Bench for UAV-Satellite Dynamic Cross-View Spatial Intelligencecited by 3
  32. 2026World2VLM: Distilling World Model Imagination into VLMs for Dynamic Spatial Reasoningcited by 3
  33. 2026ViSRA: A Video-based Spatial Reasoning Agent for Multi-modal Large Language Modelscited by 3
  34. 2026PhyWorld: Physics-Faithful World Model for Video Generationcited by 3
  35. 2026EarlyTom: Early Token Compression Completes Fast Video Understandingcited by 3
  36. 2026Watch, Remember, Reason: Human-View Video Understanding with MLLMscited by 3
  37. 2026Keep It in Mind: User Centric Continual Spatial Intelligence Reasoning in Egocentric Video Streamscited by 3
  38. 2026SPEAR: A Simulator for Photorealistic Embodied AI Researchcited by 3

50 most cited of 123 All 123 (JSON)

Similar, not cited

Close in wording, no citation either way

  1. 2026Spatial Memory Intelligence: Endowing World Models with Understanding-Driven Long-Term Memory
  2. 2024Thinking in Space: How Multimodal Large Language Models See, Remember, and Recall Spaces
  3. 2025SIMS-V: Simulated Instruction-Tuning for Spatial Video Understanding
  4. 2026SpaMEM: Benchmarking Dynamic Spatial Reasoning via Perception-Memory Integration in Embodied Environments
  5. 2026Uruqi: Learning Spatial Cognition from Visual Experience
  6. 2025UrbanVideo-Bench: Benchmarking Vision-Language Models on Embodied Intelligence with Video Data in Urban Spaces
  7. 2025Spatial Reasoning in Multimodal Large Language Models: A Survey of Tasks, Benchmarks and Methods
  8. 2026Which Pretraining Paradigm Better Serves Spatial Intelligence? An Empirical Comparison of Vision-Language and Video Generation Models