SpaCE-10: leaderboard
Each model answers the single-answer multiple-choice questions of SpaCE-10 about real indoor 3D scans, given 8 video frames; the questions combine several spatial skills (counting, size, relative position, situated observation, planning) in one item. The score is the paper's task-weighted overall accuracy, with a GPT-4o judge mapping answers to options; unlike single-skill spatial suites such as VSI-Bench, every question composes several atomic skills.
Metric: Accuracy (%). Source: arxiv.org. Saturation forecast: Around August 2027. 34 models tracked. Updated 17 Nov 2026.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | InternVL3.5-241B-A28B | 55 | #720 |
| 2 | InternVL3-38B | 53.1 | #927 |
| 3 | InternVL3-78B | 52.5 | #755 |
| 4 | GLM-4.5V | 51.6 | #758 |
| 5 | InternVL3.5-20B-A4B | 51.6 | |
| 6 | InternVL3.5-14B | 48.8 | #1503 |
| 7 | Qwen 3 VL 235B A22B Instruct | 47.9 | #540 |
| 8 | InternVL3-14B | 47.3 | #1191 |
| 9 | InternVL2.5-78B | 47.1 | #1036 |
| 10 | Claude 3.7 Sonnet | 46.2 | #476 |
| 11 | InternVL2.5-38B | 45.1 | #1366 |
| 12 | InternVL3-2B | 44.2 | #3323 |
| 13 | InternVL3.5-38B | 43.9 | #1017 |
| 14 | InternVL2.5-26B | 43.3 | #2191 |
| 15 | InternVL2.5-8B | 41.8 | #2401 |
No result here: #1 Claude Fable 5.2, #2 GPT-6.2 Astra, #3 Gemini 4 Carbon.
Interactive version: theaggregate.ai/benchmark?slug=space-10 · How It Works · Data refreshed daily, snapshot 2026-11-17.