Spatial Reasoning

November 28, 2025
by Leonardo
Benchmark and Dataset Description Number Source Illustration
MINDCUBE (Yin et al., 2025) MINDCUBE evaluates how well VLMs can build spatial mental models1 from limited views, and proposes a “map-then-reason” approach that improves VLM performance through the generation and utilization of internal structured spatial representations. 21,154 questions across 3,268 images ArkitScenes, DL3DV10K, WildRGB-D

SPAR (Zhang et al., 2025) Spatial Perception And Reasoning (SPAR) dataset is sourced from 4,500 scenes and comprises 33 spatial tasks spanning single-view, multi-view, and video settings. SPAR-7M: over 7 million QA pairs across 33 diverse spatial tasks, generated from 4,500+ richly annotated 3D indoor scenes
SPAR-Bench: 7,207 manually verified QA pairs		 ScanNet, ScanNet++, Matterport3D, Structured3D

EASI (Cai et al., 2025) Six Fundamental Capabilities: Metric Measurement (MM); Mental Reconstruction (MR); Spatial Relations (SR); Perspective-taking (PT); Deformation and Assembly (DA); Comprehensive Reasoning (CR) assembled from eight benchmark datasets, approximately 31K images, 4.5K videos, and 24K QA in total. eight benchmark datasets

VSTemporalI-Bench (Fan et al., 2025) VLMs that incorporates 3D Reconstructive instruction tuning using videos Training data: Over 200,000 general question-answer pairs for spatial reasoning from monocular video.
VSTI-Bench: approximately 138,600 QA pairs.		 ScanNet, ScanNet++, ARKitScenes

OmniSpatial (Jia et al., 2025) OmniSpatial covers four major categories: dynamic reasoning, complex spatial logic, spatial interaction, and perspective-taking, with 50 fine-grained subcategories over 8.4K question-answer pairs Internet

ViewSpatial-Bench (Li et al., 2025) Multi-viewpoint spatial localization recognition evaluation across five distinct task types, supported by an automated 3D annotation pipeline that generates precise directional labels. 5,712 question-answer pairs across 1,000+ 3D scenes MS-CoCo, ScanNet

VSI-Bench (Yang et al., 2025)

How models see, remember, and recall spaces

 5130 question-answer pairs derived from 288 real videos ScanNet, ScanNet++, ARKitScenes

OST-Bench (Lin et al., 2025) Evaluating the online spatio-temporal reasoning capabilities of MLLMs. test: 1.4k scenes and 10k question-answer pairs
train: 7k scenes and 50k question-answer pairs		 ScanNet, Matterport3D, ARKitScenes

VLM4D (Zhou et al., 2025) Evaluate the spatiotemporal reasoning capabilities 1000 videos paired with over 1800 question-answer pairs exo-centric videos: DAVIS, YouTube-VOS
ego-centric videos: Ego4D; synthetic videos: Cosmos

STARE (Li et al., 2025) Spatial Transformations and Reasoning Evaluation around 4k Objectron, HM3D

Spatial-SSRL-81k (Liu et al., 2025) A self-supervised RL paradigm that derives verifiable signals directly from ordinary RGB or RGB-D images 81,053 QA raw RGB images from COCO and RGB-D images from DIODE and MegaDepth

SpatialViz-Bench (Wang et al., 2025) A benchmark for spatial visualization with 12 tasks across 4 sub-abilities (Mental Rotation, representing and rotating two-dimensional and three-dimensional objects in space mentally while maintaining object features; 2) Mental Folding, folding two-dimensional patterns into three-dimensional objects or unfold three-dimensional objects into two-dimensional representations; 3) Visual Penetration, imagining the internal structure of objects based on external features; 4) Mental Animation) 1,180 programmatically generated problems synthetic data

MMSI-Bench (Yang et al., 2025) Multi-image spatial reasoning 1,000 multiple choice questions manually collect

LEGO-Puzzles (Tang et al., 2025) A scalable benchmark designed to evaluate both spatial understanding and sequential reasoning in MLLMs through LEGO-based tasks. 1,100 carefully curated VQA samples spanning 11 distinct tasks Internet

ERQA (Team et al., 2025) Embodied Reasoning Question Answering (ERQA) benchmark focuses specifically on capabilities likely required by an embodied agent interacting with the physical world. 400 multiple choice VQA OXE, UMI, MECCANO, HoloAssist, and EGTEA Gaze+

MultiSPA (Xu et al., 2025) Equip MLLMs with robust multi-frame spatial understanding by integrating depth perception, visual correspondence, and dynamic perception 27 million samples spanning diverse 3D and 4D scenes Aria Digital Twin (ADT), Panoptic Studio (PStudio), TAPVid3D, ScanNet

SITE (Wang et al., 2025) Spatial Intelligence Thorough Evaluation 8068 multi-choice VQA 31 computer vision datasets

3DSRBench (Ma et al., 2025) A Comprehensive 3D Spatial Reasoning Benchmark 2,772 visual question-answer pairs MS-COCO, HSSD

SpatialEval (Wang et al., 2024) Spatial-Map,Maze-Nav, Spatial-Grid, and Spatial-Real Text-only (TQA): 4640 Vision-only (VQA): 4640 Vision-Text (VTQA): 4640 synthetic, Densely Captioned Images (DCI)

DSI-Bench (Zhang et al., 2025) Dynamic Spatial Intelligence nearly 1,000 dynamic videos and over 1,700 manually annotated questions Kinetics-700, the synthetic motion-control dataset, LLaVA-178K, additional online sources.

VST (Yang et al., 2025) Visual Spatial Tuning VST-Perception (VST-P): 4.1 M samples across 19 different tasks; VST-Reasoning (VST-R): 135K samples open-source datasets

SAT (Ray et al., 2025) A simulated spatial aptitude training dataset comprising both static and dynamic spatial reasoning 175K question-answer (QA) pairs and 20K scenes ProcTHOR-10K

CoVT (Qin et al., 2025) Enables VLMs to reason not only in words but also through continuous visual tokens 774.6k LLaVA-OneVision, TallyQA and ADE20K-Depth

MSMU (Chen et al., 2025) Massive Spatial Measuring and Understanding (MSMU) dataset with precise spatial annotations 700K QA pairs, 2.5M physical numerical annotations, and 10K chain-of-thought augmented samples ScanNet, ScanNet++

What’sUp (Kamath et al., 2023) Contains sets of photographs varying only the spatial relations of objects, keeping their identity fixed 4,138 COCO,GQA

RoboSpatial (Song et al., 2025) A large-scale dataset for spatial understanding in robotics 1M images, 5k 3D scans, and 3M annotated spatial relationships Matterport3D, ScanNet, 3RScan, HOPE, GraspNet-1B

Spatial-MM (Shiri et al., 2024) Benchmark for spatial understanding and reasoning capabilities Spatial-Obj: 2,000 multiple-choice questions; Spatial-CoT: 310 spatial-aware multi-hop QA pairs Internet

Super-CLEVR-3D (Wang et al., 2023) 3D-aware VQA, which focuses on challenging questions that require a compositional reasoning over the 3D structure of visual scenes 30k images Super-CLEVR

SpatialQA (Cai et al., 2024) Multi-level depth related questions spanning various scenarios and embodiment tasks 852,869 Bunny 695k, Open X-Embodiment

SURDS (Guo et al., 2024) Benchmarking Spatial Understanding and Reasoning in Driving Scenarios 41,080 vision–question–answer training instances and 9,250 evaluation samples nuScenes

Open3D-VQA (Zhang et al., 2025) Evaluating MLLMs’ ability to reason about complex spatial relationships from an aerial perspective 73k QA pairs spanning 7 general spatial reasoning tasks synthetic data

RefSpatial (Zhou et al., 2025) Spatial referring 2.5M high-quality examples with 20M QA pairs (2× prior), covering 31 spatial relations (vs. 15 prior) and supporting complex reasoning processes (up to 5 steps OpenImages, CA-1M, Infinigen

SSR-COT (Liu et al., 2025) A million-scale visual-language reasoning dataset enriched with intermediate spatial reasoning annotations 1.2M LLaVA-CoT, Visual-CoT, VoCoT, SpatialQA

SpaceR-91k (Ouyang et al., 2025) A tailored spatial reasoning dataset built upon the 3D indoor scene reconstruction dataset ScanNet 91k questions spanning diverse spatial reasoning scenarios with verifiable answers ScanNet

MulSeT (Zhang et al., 2025) Multi-view Spatial Understanding Tasks 38.2k question-answer pairs spanning more than 5,000 unique 3D scenes AI2THOR

Spartun3D (Zhang et al., 2024) Incorporates various situated spatial reasoning tasks 10k situated captions and 123k QA pairs. 3RScan

MSQA (Linghu et al., 2024) 3D situated reasoning 251K situated QA pairs ScanNet, 3RScan, ARKitScenes

SQA3D (Ma et al., 2022) Situated Question Answering in 3D Scenes 33.4k diverse reasoning questions ScanNet

VQASynth (remyxai, 2024) Compose multimodal datasets - Internet

AS-V2 (Wang et al., 2024) High-quality ReC dataset 127k COCO

OSD (Cheng et al., 2024) Open Spatial Dataset 8.7M spatial concepts grounded in 5M unique regions from 1M images OpenImages

Proximity-110K (Li et al., 2024) Proximity Question Answering 559,952 Q-A pairs related to object depth information and 429,925 Q-A pairs focusing on object proximity relationships Visual Genome, COCO

TopViewRS (Li et al., 2024) Top-View Reasoning in Space 11,384 multiple-choice questions Matterport3D

EmbSpatial-Bench (Du et al., 2024) Evaluating embodied spatial understanding of LVLMs. 3,640 MP3D, AI2-THOR, ScanNet

SpaCE-10 (Gong et al., 2025) Compositional spatial evaluations 5k+ QA pairs in 811 indoor scenes SCN, 3RS, ARK, SCN++

Ego3D-Bench (Gholami et al., 2025) Evaluate the spatial reasoning abilities of VLMs using ego-centric, multi-view outdoor data over 8,600 QA pairs manually collect

SpatialScore Comprehensive and diverse multimodal spatial understanding benchmark 28K samples spanning 8 categories MMVP, RealWorldQA, SpatialSense, SpatialBench, QSpatialBench, CV-Bench, VSR, 3DSRBench, VSI-Bench, BLINK, MMIU

All-Angles Bench (Yeh et al., 2025) Multi-view scene reasoning 2,132 question–answer pairs Exo4D, EgoHumans

STI-Bench (Li et al., 2025) Evaluate models’ Spatial-Temporal Intelligence 2,060 Waymo, ScanNet, Omni6DPose

SPACE (Ramakrishnan et al., 2025) Evaluates spatial cognition 5,008 Synthetic Data

Q-Spatial Bench (Liao et al., 2024) Quantitative spatial reasoning 271 ScanNet, images captured by iPhone

SenseNova-SI-8M (Cai et al., 2025) SenseNova-SI-8M reorganizes 4M open-source data and scales 4.5M additional data, according to fundamental spatial capbilities 8.5M QA pairs existing open-source datasets and synthetic data

Viewpoint-100K (Zhan et al., 2025) Viewpoint Learning 100K object-centric image pairs and the corresponding QAs MVImgNet

H∗Bench (Yu et al., 2025) Humanoid visual search: A humanoid agent actively rotates its head to search for objects or paths in an immersive world represented by a 360° panoramic image approximately 3,000 annotated task instances self-collected footage across global metropolitan areas (New York, Paris, Amsterdam, Frankfurt) and open platforms (YouTube and the 360+x datase

Puffin-4M (Liao et al., 2025) vision-language-camera triplets 4M existing open-source datasets and online

TIGeR-300K (Han et al., 2025) tool-invocation-oriented dataset 300k CA-1M

    1. Spatial mental model: an internal representation of the environment that allows for consistent understanding and inference about space, independent of the current viewpoint.

References

  1. Ray, A., Duan, J., II, E. L. B., Tan, R., Bashkirova, D., Hendrix, R., Ehsani, K., Kembhavi, A., Plummer, B. A., Krishna, R., Zeng, K.-H., and Saenko, K. SAT: Dynamic Spatial Aptitude Training for Multimodal Language Models. Second Conference on Language Modeling, 2025.
  2. Gholami, M., Rezaei, A., Weimin, Z., Mao, S., Zhou, S., Zhang, Y., and Akbari, M. Spatial Reasoning with Vision-Language Models in Ego-Centric Multi-View Scenes. Arxiv Preprint Arxiv: 2509.06266, 2025.
  3. Du, M., Wu, B., Li, Z., Huang, X., and Wei, Z. EmbSpatial-Bench: Benchmarking Spatial Understanding for Embodied Tasks with Large Vision-Language Models. Arxiv Preprint Arxiv: 2406.05756, 2024.
  4. Zhan, X., Huang, W., Sun, H., Fu, X., Ma, C., Cao, S., Jia, B., Lin, S., Yin, Z., Bai, L., Ouyang, W., Li, Y., Guo, J., and Guo, Y. Actial: Activate Spatial Reasoning Ability of Multimodal Large Language Models. Arxiv Preprint Arxiv: 2511.01618, 2025.
  5. Yu, H., Han, Y., Zhang, X., Yin, B., Chang, B., Han, X., Liu, X., Zhang, J., Pavone, M., Feng, C., Xie, S., and Li, Y. Thinking in 360\textdegree{}: Humanoid Visual Search in the Wild. Arxiv Preprint Arxiv: 2511.20351, 2025.
  6. Hu, W., Lin, J., Long, Y., Ran, Y., Jiang, L., Wang, Y., Zhu, C., Xu, R., Wang, T., and Pang, J. G^2VLM: Geometry Grounded Vision Language Model with Unified 3D Reconstruction and Spatial Reasoning. Arxiv Preprint Arxiv: 2511.21688, 2025.
  7. remyxai. VQASynth, 2024.
  8. Tang, K., Gao, J., Zeng, Y., Duan, H., Sun, Y., Xing, Z., Liu, W., Lyu, K., and Chen, K. LEGO-Puzzles: How Good Are MLLMs at Multi-Step Spatial Reasoning?. Arxiv Preprint Arxiv: 2503.19990, 2025.
  9. Zhang, W., Zhou, Z., Zeng, X., Liu, X., Fang, J., Gao, C., Li, Y., Cui, J., Chen, X., and Zhang, X.-P. Open3D-VQA: A Benchmark for Comprehensive Spatial Reasoning with Multimodal Large Language Model in Open Space. Arxiv Preprint Arxiv: 2503.11094, 2025.
  10. Lin, J., Zhu, C., Xu, R., Mao, X., Liu, X., Wang, T., and Pang, J. OST-Bench: Evaluating the Capabilities of MLLMs in Online Spatio-temporal Scene Understanding. Arxiv Preprint Arxiv: 2507.07984, 2025.
  11. Ma, W., Chen, H., Zhang, G., Chou, Y.-C., Chen, J., Melo, C. de, and Yuille, A. 3DSRBench: A Comprehensive 3D Spatial Reasoning Benchmark. Proceedings of the IEEE/CVF International Conference on Computer Vision (ICCV), 2025.
  12. Zhou, E., An, J., Chi, C., Han, Y., Rong, S., Zhang, C., Wang, P., Wang, Z., Huang, T., Sheng, L., and Zhang, S. RoboRefer: Towards Spatial Referring with Reasoning in Vision-Language Models for Robotics. Arxiv Preprint Arxiv: 2506.04308, 2025.
  13. Yang, J., Yang, S., Gupta, A. W., Han, R., Fei-Fei, L., and Xie, S. Thinking in Space: How Multimodal Large Language Models See, Remember, and Recall Spaces. Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), 2025.
  14. Gong, Z., Li, W., Ma, O., Li, S., Wang, Z., Li, S., Ji, J., Yang, X., Luo, G., Yan, J., and Ji, R. SpaCE-10: A Comprehensive Benchmark for Multimodal Large Language Models in Compositional Spatial Intelligence. Arxiv Preprint Arxiv: 2506.07966, 2025.
  15. Li, Y., Zhang, Y., Lin, T., Liu, X., Cai, W., Liu, Z., and Zhao, B. STI-Bench: Are MLLMs Ready for Precise Spatial-Temporal World Understanding?. Arxiv Preprint Arxiv: 2503.23765, 2025.
  16. Cheng, A.-C., Yin, H., Fu, Y., Guo, Q., Yang, R., Kautz, J., Wang, X., and Liu, S. SpatialRGPT: Grounded Spatial Reasoning in Vision-Language Models. Advances in Neural Information Processing Systems, 2024.
  17. Liao, K., Wu, S., Wu, Z., Jin, L., Wang, C., Wang, Y., Wang, F., Li, W., and Loy, C. C. Thinking with Camera: A Unified Multimodal Model for Camera-Centric Understanding and Generation. Arxiv Preprint Arxiv: 2510.08673, 2025.
  18. Zhang, Y., Xu, Z., Shen, Y., Kordjamshidi, P., and Huang, L. SPARTUN3D: Situated Spatial Understanding of 3D World in Large Language Models. International Conference on Learning Representations, 2024.
  19. Guo, X., Zhang, R., Duan, Y., He, Y., Nie, D., Huang, W., Zhang, C., Liu, S., Zhao, H., and Chen, L. SURDS: Benchmarking Spatial Understanding and Reasoning in Driving Scenarios with Vision Language Models. Arxiv Preprint Arxiv: 2411.13112, 2024.
  20. Liao, Y.-H., Mahmood, R., Fidler, S., and Acuna, D. Reasoning Paths with Reference Objects Elicit Quantitative Spatial Reasoning in Large Vision-Language Models. Arxiv Preprint Arxiv: 2409.09788, 2024.
  21. Zhou, S., Vilesov, A., He, X., Wan, Z., Zhang, S., Nagachandra, A., Chang, D., Chen, D., Wang, X. E., and Kadambi, A. VLM4D: Towards Spatiotemporal Awareness in Vision Language Models. Proceedings of the IEEE/CVF International Conference on Computer Vision (ICCV), 2025.
  22. Li, L., Bigverdi, M., Gu, J., Ma, Z., Yang, Y., Li, Z., Choi, Y., and Krishna, R. Unfolding Spatial Cognition: Evaluating Multimodal Models on Visual Simulations. Arxiv Preprint Arxiv: 2506.04633, 2025.
  23. Zhang, W., Huang, Y., Xu, Y., Huang, J., Zhi, H., Ren, S., Xu, W., and Zhang, J. Why Do MLLMs Struggle with Spatial Understanding? A Systematic Analysis from Data to Architecture. Arxiv Preprint Arxiv: 2509.02359, 2025.
  24. Zhang, J., Chen, Y., Zhou, Y., Xu, Y., Huang, Z., Mei, J., Chen, J., Yuan, Y.-J., Cai, X., Huang, G., Quan, X., Xu, H., and Zhang, L. From Flatland to Space: Teaching Vision-Language Models to Perceive and Reason in 3D. Arxiv Preprint Arxiv: 2503.22976, 2025.
  25. Liu, Y., Zhang, B., Zang, Y., Cao, Y., Xing, L., Dong, X., Duan, H., Lin, D., and Wang, J. Spatial-SSRL: Enhancing Spatial Understanding via Self-Supervised Reinforcement Learning. Arxiv Preprint Arxiv: 2510.27606, 2025.
  26. Wang, W., Ren, Y., Luo, H., Li, T., Yan, C., Chen, Z., Wang, W., Li, Q., Lu, L., Zhu, X., Qiao, Y., and Dai, J. The All-Seeing Project V2: Towards General Relation Comprehension of the Open World. Computer Vision - ECCV 2024 - 18th European Conference, Milan, Italy, September 29-October 4, 2024, Proceedings, Part XXXIII, 2024.
  27. Wang, W., Tan, R., Zhu, P., Yang, J., Yang, Z., Wang, L., Kolobov, A., Gao, J., and Gong, B. SITE: towards Spatial Intelligence Thorough Evaluation. Arxiv Preprint Arxiv: 2505.05456, 2025.
  28. Gupta, A., DollĂĄr, P., and Girshick, R. LVIS: A Dataset for Large Vocabulary Instance Segmentation. Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), 2019.
  29. Han, Y., Chi, C., Zhou, E., Rong, S., An, J., Wang, P., Wang, Z., Sheng, L., and Zhang, S. TIGeR: Tool-Integrated Geometric Reasoning in Vision-Language Models for Robotics. Arxiv Preprint Arxiv: 2510.07181, 2025.
  30. Li, C., Zhang, C., Zhou, H., Collier, N., Korhonen, A., and Vuli'c, I. TopViewRS: Vision-Language Models as Top-View Spatial Reasoners. Conference on Empirical Methods in Natural Language Processing, 2024.
  31. Yang, S., Xu, R., Xie, Y., Yang, S., Li, M., Lin, J., Zhu, C., Chen, X., Duan, H., Yue, X., Lin, D., Wang, T., and Pang, J. MMSI-Bench: A Benchmark for Multi-Image Spatial Intelligence. Arxiv Preprint Arxiv: 2505.23764, 2025.
  32. Qin, Y., Wei, B., Ge, J., Kallidromitis, K., Fu, S., Darrell, T., and Wang, X. Chain-of-Visual-Thought: Teaching VLMs to See and Think Better with Continuous Visual Tokens. Arxiv Preprint Arxiv: 2511.19418, 2025.
  33. Yang, R., Zhu, Z., Li, Y., Huang, J., Yan, S., Zhou, S., Liu, Z., Li, X., Li, S., Wang, W., Lin, Y., and Zhao, H. Visual Spatial Tuning. Arxiv Preprint Arxiv: 2511.05491, 2025.
  34. Jia, M., Qi, Z., Zhang, S., Zhang, W., Yu, X., He, J., Wang, H., and Yi, L. OmniSpatial: Towards Comprehensive Spatial Reasoning Benchmark for Vision Language Models. Arxiv Preprint Arxiv: 2506.03135, 2025.
  35. Kamath, A., Hessel, J., and Chang, K.-W. What's "up" with vision-language models? Investigating their struggle with spatial reasoning. Conference on Empirical Methods in Natural Language Processing, 2023.
  36. Wang, X., Ma, W., Li, Z., Kortylewski, A., and Yuille, A. L. 3D-Aware Visual Question Answering about Parts, Poses and Occlusions. Advances in Neural Information Processing Systems, 2023.
  37. Chen, P., Lou, Y., Cao, S., Guo, J., Fan, L., Wu, Y., Yang, L., Ma, L., and Ye, J. SD-VLM: Spatial Measuring and Understanding with Depth-Encoded Vision-Language Models. Arxiv Preprint Arxiv:2509.17664, 2025.
  38. Ma, X., Yong, S., Zheng, Z., Li, Q., Liang, Y., Zhu, S.-C., and Huang, S. SQA3D: Situated Question Answering in 3D Scenes. International Conference on Learning Representations, 2022.
  39. Li, P., Zhang, H., Sun, M., Shao, R., Xiang, T., Wang, W., and Shan, Y. MVImgNet: A Large-scale Dataset of Multi-view Images. Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), 2023.
  40. Wang, J., Ming, Y., Shi, Z., Vineet, V., Wang, X., Li, Y., and Joshi, N. Is A Picture Worth A Thousand Words? Delving Into Spatial Reasoning for Vision Language Models. Advances in Neural Information Processing Systems, 2024.
  41. Xu, R., Wang, W., Tang, H., Chen, X., Wang, X., Chu, F.-J., Lin, D., Feiszli, M., and Liang, K. J. Multi-SpatialMLLM: Multi-Frame Spatial Understanding with Multi-Modal Large Language Models. Arxiv Preprint Arxiv: 2505.17015, 2025.
  42. Shiri, F., Guo, X.-Y., Far, M. G., Yu, X., Haf, R., and Li, Y.-F. An Empirical Analysis on Spatial Reasoning Capabilities of Large Multimodal Models. Proceedings of the 2024 Conference on Empirical Methods in Natural Language Processing, 2024.
  43. Linghu, X., Huang, J., Niu, X., Ma, X., Jia, B., and Huang, S. Multi-modal Situated Reasoning in 3D Scenes. Advances in Neural Information Processing Systems, 2024.
  44. Yin, B., Wang, Q., Zhang, P., Zhang, J., Wang, K., Wang, Z., Zhang, J., Chandrasegaran, K., Liu, H., Krishna, R., Xie, S., Li, M., Wu, J., and Fei-Fei, L. Spatial Mental Modeling from Limited Views. Structural Priors for Vision Workshop at Iccv'25, 2025.
  45. Ramakrishnan, S. K., Wijmans, E., Krähenbßhl, P., and Koltun, V. Does Spatial Cognition Emerge in Frontier Models?. The Thirteenth International Conference on Learning Representations, ICLR 2025, Singapore, April 24-28, 2025, 2025.
  46. Li, J., Nan, X., Lu, M., Du, L., and Zhang, S. Proximity QA: Unleashing the Power of Multi-Modal Large Language Models for Spatial Proximity Analysis. Arxiv Preprint Arxiv: 2401.17862, 2024.
  47. Cai, W., Ponomarenko, I., Yuan, J., Li, X., Yang, W., Dong, H., and Zhao, B. SpatialBot: Precise Spatial Understanding with Vision Language Models. Arxiv Preprint Arxiv: 2406.13642, 2024.
  48. Team, G. R., Abeyruwan, S., Ainslie, J., Alayrac, J.-B., Arenas, M. G., Armstrong, T., Balakrishna, A., Baruch, R., Bauza, M., Blokzijl, M., Bohez, S., Bousmalis, K., Brohan, A., Buschmann, T., Byravan, A., Cabi, S., Caluwaerts, K., Casarini, F., Chang, O., … Zhou, Y. Gemini Robotics: Bringing AI into the Physical World. Arxiv Preprint Arxiv: 2503.20020, 2025.
  49. Li, D., Li, H., Wang, Z., Yan, Y., Zhang, H., Chen, S., Hou, G., Jiang, S., Zhang, W., Shen, Y., Lu, W., and Zhuang, Y. ViewSpatial-Bench: Evaluating Multi-perspective Spatial Localization in Vision-Language Models. Arxiv Preprint Arxiv: 2505.21500, 2025.
  50. Cai, Z., Wang, R., Gu, C., Pu, F., Xu, J., Wang, Y., Yin, W., Yang, Z., Wei, C., Sun, Q., Zhou, T., Li, J., Pang, H. E., Qian, O., Wei, Y., Lin, Z., Shi, X., Deng, K., Han, X., … Yang, L. Scaling Spatial Intelligence with Multimodal Foundation Models. Arxiv Preprint Arxiv: 2511.13719, 2025.
  51. Fan, Z., Zhang, J., Li, R., Zhang, J., Chen, R., Hu, H., Wang, K., Qu, H., Wang, D., Yan, Z., Xu, H., Theiss, J., Chen, T., Li, J., Tu, Z., Wang, Z., and Ranjan, R. VLM-3R: Vision-Language Models Augmented with Instruction-Aligned 3D Reconstruction. Arxiv Preprint Arxiv: 2505.20279, 2025.
  52. Yeh, C.-H., Wang, C., Tong, S., Cheng, T.-Y., Wang, R., Chu, T., Zhai, Y., Chen, Y., Gao, S., and Ma, Y. Seeing from Another Perspective: Evaluating Multi-View Understanding in MLLMs. Arxiv Preprint Arxiv: 2504.15280, 2025.
  53. Zhang, Z., Wang, Z., Zhang, G., Dai, W., Xia, Y., Yan, Z., Hong, M., and Zhao, Z. DSI-Bench: A Benchmark for Dynamic Spatial Intelligence. Arxiv Preprint Arxiv: 2510.18873, 2025.
  54. Cai, Z., Wang, Y., Sun, Q., Wang, R., Gu, C., Yin, W., Lin, Z., Yang, Z., Wei, C., Qian, O., Pang, H. E., Shi, X., Deng, K., Han, X., Chen, Z., Li, J., Fan, X., Deng, H., Lu, L., … Yang, L. Holistic Evaluation of Multimodal LLMs on Spatial Intelligence. Arxiv Preprint Arxiv: 2508.13142, 2025.
  55. Ouyang, K., Liu, Y., Wu, H., Liu, Y., Zhou, H., Zhou, J., Meng, F., and Sun, X. SpaceR: Reinforcing MLLMs in Video Spatial Reasoning. Arxiv Preprint Arxiv: 2504.01805, 2025.
  56. Wang, S., Pei, M., Sun, L., Deng, C., Shao, K., Tian, Z., Zhang, H., and Wang, J. SpatialViz-Bench: An MLLM Benchmark for Spatial Visualization. Arxiv Preprint Arxiv: 2507.07610, 2025.
  57. Song, C. H., Blukis, V., Tremblay, J., Tyree, S., Su, Y., and Birchfield, S. RoboSpatial: Teaching Spatial Understanding to 2D and 3D Vision-Language Models for Robotics. Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), 2025.
  58. Liu, Y., Ma, M., Yu, X., Ding, P., Zhao, H., Sun, M., Huang, S., and Wang, D. SSR: Enhancing Depth Perception in Vision-Language Models via Rationale-Guided Spatial Reasoning. Arxiv Preprint Arxiv: 2505.12448, 2025.
  1. 1Spatial mental model: an internal representation of the environment that allows for consistent understanding and inference about space, independent of the current viewpoint.