Uruqi: leaderboard
Spatial cognition from continuous visual experience: over 52k questions on 2.7k simulated camera-trajectory episodes, a vision-language model must track its own motion, keep a persistent map of the objects it has seen and apply spatial state operations (updates, relations, events) as the view changes. Scored by overall accuracy, the mean of the three stage scores; higher is better (Uruqi, 2026).
Metric: Overall accuracy (%): mean of the three stage scores (self-motion, object mapping, state operations), subclasses weighted equally within each stage, over 52k questions on 2.7k simulated camera-trajectory episodes; higher is better. Source: arxiv.org. Saturation forecast: Around July 2028. 12 models tracked. Updated 2 Oct 2026.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | GPT-6 Astra | 50.08 | #8 |
| 2 | Gemini 3.1 Pro (Preview) | 27.37 | #99 |
| 3 | GPT-5.5 | 24.5 | #50 |
| 4 | Qwen 3.8 Max | 23.58 | #64 |
| 5 | Qwen 3 VL 8B | 18.61 | #919 |
| 6 | SenseNova-SI-1.5-8B | 17.02 | |
| 7 | Qwen 3.8 27B | 16.94 | #219 |
| 8 | InternVL3-8B | 15.84 | #1756 |
| 9 | SpaceR-7B | 13.64 | #1169 |
| 10 | VST-7B-SFT | 12.99 | #1001 |
| 11 | MindCube-RL | 10.37 | |
| 12 | Spatial-MLLM | 4.99 |
No result here: #1 Claude Fable 5.2, #2 GPT-6.2 Astra, #3 Gemini 4 Carbon.
Interactive version: theaggregate.ai/benchmark?slug=uruqi · How It Works · Data refreshed daily, snapshot 2026-11-17.