PanoSpace-Bench - Object Reorientation: leaderboard
Metric: Accuracy (%; 250 five-option multiple-choice questions on one ERP panorama). Source: arxiv.org. Saturation forecast: Around October 2027. 12 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | MiMo-V2.5 | 42 |
| 2 | Qwen 2.5 VL 7B Instruct | 30.4 |
| 3 | Qwen 3 VL 8B | 27.2 |
| 4 | Qwen 3.5 9B | 26.3 |
| 5 | GPT-4o | 24.8 |
| 6 | Ministral 3 8B | 20 |
| 7 | Ministral 3 14B | 12.8 |
Interactive version: theaggregate.ai/benchmark?slug=panospace-bench-object-reorientation · How It Works · Data refreshed daily, snapshot 2026-09-25.