VSI-Bench
InternVL3.5-241B-A28B 69.5% leads since Aug 2025 Fading Models watch egocentric videos of 288 indoor scenes and answer over 5,000 questions on object counts, distances, sizes, room layout and route planning.
Mixes multiple-choice questions scored by accuracy with numeric answers scored by mean relative accuracy. Videos come from ScanNet, ScanNet++ and ARKitScenes.
In the authors’ words
Humans possess the visual-spatial intelligence to remember spaces from sequential visual observations. However, can Multimodal Large Language Models (MLLMs) trained on million-scale video datasets also “think in space” from videos? We present a novel video-based visual-spatial intelligence benchmark (VSI-Bench) of over 5,000 question-answer pairs, and find that MLLMs exhibit competitive - though subhuman - visual-spatial intelligence. We probe models to express how they think in space both linguistically and visually and find that while spatial reasoning capabilities remain the primary bottleneck for MLLMs to reach higher benchmark performance, local world models and spatial awareness do emerge within these models. Notably, prevailing linguistic reasoning techniques (e.g., chain-of-thought, self-consistency, tree-of-thoughts) fail to improve performance, whereas explicitly generating cognitive maps during question-answering enhances MLLMs' spatial distance ability.
- Level
- Everyday Most adults, without training
- Scored
- Checked automatically A program checks the result against an answer key, unit tests or the expected final state.
- Answers
- Multiple choice
InternVL3.5-241B-A28B leads with 69.5% among models released by Jan 2026
Best score per model as labs report them in their own cards. Labs pick their own settings, so these compare less cleanly than independent runs.
The newest model with a score here came out in Jan 2026, and nobody has reported a newer one. Among the 20 highest-ranked models on the capability index, these have no score yet:
- GPT-6 Astra
- Claude Opus 5.5
- GPT-6.1 Terra
- GPT-6.1 Sol
- and 16 more
- 1 InternVL3.5-241B-A28B 69.5%
- 2
SenseNova-SI-1.3-InternVL3-8B 68.6% - 3
Qwen3-VL-235B-A22B 60% - 4
gemini-robotics-er-1.5-preview 45.8%
| # | Model | Score | Setting | Date | Source |
|---|---|---|---|---|---|
| 1 | InternVL3.5-241B-A28B | 69.5% | Aug 2025 | Model card | |
| 2 | SenseNova-SI-1.3-InternVL3-8B | 68.6% | Jan 2026 | Model card | |
| 3 | Qwen3-VL-235B-A22B | 60% | Nov 2025 | Model card | |
| 4 | gemini-robotics-er-1.5-preview | 45.8% | Sep 2025 | Model card |
Number one over time
Lab claims count from the day the card came out, independent runs from the model’s release.
Scores over time
All the scores we have, one dot each, so a model tested at several settings appears more than once.
Labs headline it less often now, last in Jan 2026
Launch posts from every lab that has listed it, one tick per post. Tall colored ticks include it, and short gray ones leave it out.
| Date | Launch post | Score | As the post writes it |
|---|---|---|---|
| 8 Jan 2026 | SenseNova-SI-1.3-InternVL3-8B | 68.6 | VSI |
| 25 Aug 2025 | InternVL3.5-241B-A28B | 69.5 | VSI-Bench |
3 system cards report it, from 3 labs
Besides launch posts, we read the system cards, model cards and technical reports labs publish with a release. They report far more, including safety and frontier-risk tests.
| Date | Document | Score | As the document writes it | Section |
|---|---|---|---|---|
| 26 Nov 2025 | Qwen3-VL-235B-A22B technical report | 60 | VSI-Bench (Table 2) | Capabilities |
| 25 Sep 2025 | gemini-robotics-er-1.5-preview technical report | 45.8 | VSI-Bench (Table 19, thinking on). Thinking off 39.9 | Capabilities |
| 25 Aug 2025 | InternVL3.5-241B-A28B technical report | 69.5 | VSI-Bench (Table 2, Table 11) | Capabilities |
More in Vision
- MMMU In use 11.5K college-level multimodal questions across 30 subjects. Multiple-choice and open accuracy. GPT-5.1 85.4%
- MMMU-Pro In use Harder MMMU: college-level multimodal questions with more options and vision-only inputs to remove text shortcuts. Kimi K3.1 Pro 84.1%
- CharXiv In use Chart understanding on real arXiv figures. Reasoning questions require synthesizing information from complex charts. Claude Fable 5 93.5%
- MATH-Vision (MathVision) In use 3,040 competition math problems with visual context. Answer accuracy. Qwen 3.8 Max 97.7%
- ChartQA In use Models answer human-written and generated questions about chart images that need visual and arithmetic reasoning. Scored by relaxed accuracy with a 5 percent numeric tolerance.
Command A Vision 90.9% - DocVQA In use Models answer 50,000 questions about 12,000 scanned document images by reading their text and layout. Scoring is ANLS, a string similarity to reference answers.
Seed1.5-VL 96.9%