# ReVSI: Rebuilding Visual Spatial Intelligence Evaluation for Accurate Assessment of VLM 3D Reasoning

arXiv: https://arxiv.org/abs/2604.24300 · HTML: https://mlsanity.com/paper/2604.24300/

## Key facts

- Cited by 15 papers in the map (data through 14 Oct 2026).
- Placed in Spatial and embodied reasoning, one of 17 directions in LLM reasoning.
- Builds on 18 of its 34 references that are in the map.
- First posted on arXiv 27 Apr 2026.

## Placed in

- LLM reasoning / [Spatial and embodied reasoning](https://mlsanity.com/lr/spatial-and-embodied-reasoning/)
- Multimodal LLMs / vision-language models / [Embodied and 3D vision-language models](https://mlsanity.com/mllm/embodied-and-3d-vision-language-models/)

## Builds on (most cited first, 18 of 18)

- [Qwen3-VL Technical Report](https://mlsanity.com/paper/2511.21631/) (2025)
- [InternVL3.5: Advancing Open-Source Multimodal Models in Versatility, Reasoning, and Efficiency](https://mlsanity.com/paper/2508.18265/) (2025)
- [Gemini: A Family of Highly Capable Multimodal Models](https://mlsanity.com/paper/2312.11805/) (2023)
- [Open3D: A Modern Library for 3D Data Processing](https://mlsanity.com/paper/1801.09847/) (2018)
- [SWIFT:A Scalable lightWeight Infrastructure for Fine-Tuning](https://mlsanity.com/paper/2408.05517/) (2024)
- [LMMs-Eval: Reality Check on the Evaluation of Large Multimodal Models](https://mlsanity.com/paper/2407.12772/) (2024)
- [Spatial-MLLM: Boosting MLLM Capabilities in Visual-based Spatial Intelligence](https://mlsanity.com/paper/2505.23747/) (2025)
- [VLM-3R: Vision-Language Models Augmented with Instruction-Aligned 3D Reconstruction](https://mlsanity.com/paper/2505.20279/) (2025)
- [SpaceR: Reinforcing MLLMs in Video Spatial Reasoning](https://mlsanity.com/paper/2504.01805/) (2025)
- [From Flatland to Space: Teaching Vision-Language Models to Perceive and Reason in 3D](https://mlsanity.com/paper/2503.22976/) (2025)
- [Reinforcing Spatial Reasoning in Vision-Language Models with Interwoven Thinking and Visual Drawing](https://mlsanity.com/paper/2506.09965/) (2025)
- [Cambrian-S: Towards Spatial Supersensing in Video](https://mlsanity.com/paper/2511.04670/) (2025)
- [RoboRefer: Towards Spatial Referring with Reasoning in Vision-Language Models for Robotics](https://mlsanity.com/paper/2506.04308/) (2025)
- [EmbSpatial-Bench: Benchmarking Spatial Understanding for Embodied Tasks with Large Vision-Language Models](https://mlsanity.com/paper/2406.05756/) (2024)
- [Visual Spatial Tuning](https://mlsanity.com/paper/2511.05491/) (2025)
- [3D-R1: Enhancing Reasoning in 3D VLMs for Unified Scene Understanding](https://mlsanity.com/paper/2507.23478/) (2025)
- [Beyond Flatlands: Unlocking Spatial Intelligence by Decoupling 3D Reasoning from Numerical Regression](https://mlsanity.com/paper/2511.11239/) (2025)
- [Video Spatial Reasoning with Object-Centric 3D Rollout](https://mlsanity.com/paper/2511.13190/) (2025)

## Built on by (most cited first, 15 of 15)

- [LLaVA-OneVision-2: Towards Next-Generation Perceptual Intelligence](https://mlsanity.com/paper/2605.25979/) (2026)
- [S-Agent: Spatial Tool-Use Elicits Reasoning for Spatial Intelligence](https://mlsanity.com/paper/2606.20515/) (2026)
- [Cambrian-P: Pose-Grounded Video Understanding](https://mlsanity.com/paper/2605.22819/) (2026)
- [Annotations as Rollouts: Efficient and Scalable Reinforcement Learning for Video MLLMs](https://mlsanity.com/paper/2608.20492/) (2026)
- [ReSTI: A Source-Grounded Audit and Repair of STI-Bench](https://mlsanity.com/paper/2609.24727/) (2026)
- [Skill-3D: Evolving Scene-Aware Skills for Agentic 3D Spatial Reasoning](https://mlsanity.com/paper/2606.07436/) (2026)
- [3DHarnessBench: Probing Agentic 3D-to-Code Capabilities of Frontier Vision-Language Models](https://mlsanity.com/paper/2609.06535/) (2026)
- [GeoWeaver: Grounding Visual Tokens with Geometric Evidence before Scene Reasoning](https://mlsanity.com/paper/2605.22558/) (2026)
- [Seeing Is Not Measuring: Tool-Augmented Metric Spatial Reasoning for Vision-Language Models](https://mlsanity.com/paper/2609.29073/) (2026)
- [SpatialSpeak: QA-Native Reconstruction with Local and Global Context for Spatial Chain-of-Thought Reasoning](https://mlsanity.com/paper/2609.33616/) (2026)
- [From Reasoning Failures to Composable Video Spatial Intelligence](https://mlsanity.com/paper/2610.01999/) (2026)
- [Render to Reason: Novel-View Semantic Prediction Improves Spatial Understanding in VLMs](https://mlsanity.com/paper/2610.05417/) (2026)
- [MemAct: Streaming Spatial Understanding with Memory and Active Exploration](https://mlsanity.com/paper/2610.17640/) (2026)
- [GeoThinker-F: Endogenous Predictive World Representations for Fast Spatial Intelligence](https://mlsanity.com/paper/2610.20592/) (2026)
- [DynThinker: Unifying Static 3D and Dynamic 4D Spatial Reasoning](https://mlsanity.com/paper/2610.20598/) (2026)

## Abstract (arXiv)

Current evaluations of spatial intelligence can be systematically invalid under modern vision-language model (VLM) settings. First, many benchmarks derive question-answer (QA) pairs from point-cloud-based 3D annotations originally curated for traditional 3D perception. When such annotations are treated as ground truth for video-based evaluation, reconstruction and annotation artifacts can miss objects that are clearly visible in the video, mislabel object identities, or corrupt geometry-dependent answers (e.g., size), yielding incorrect or ambiguous QA pairs. Second, evaluations often assume full-scene access, while many VLMs operate on sparsely sampled frames (e.g., 16-64), making many questions effectively unanswerable under the actual model inputs. We improve evaluation validity by introducing ReVSI, a benchmark and protocol that ensures each QA pair is answerable and correct under the model's actual inputs. To this end, we re-annotate objects and geometry across 381 scenes from 5 datasets to improve data quality, and regenerate all QA pairs with rigorous bias mitigation and human verification using professional 3D annotation tools. We further enhance evaluation controllability by providing variants across multiple frame budgets (16/32/64/all) and fine-grained object visibility metadata, enabling controlled diagnostic analyses. Evaluations of general and domain-specific VLMs on ReVSI reveal systematic failure modes that are obscured by prior benchmarks, yielding a more reliable and diagnostic assessment of spatial intelligence.