Embodied3DBench - Spatial Relation (3D): leaderboard
Metric: Spatial Relation (3D) task (object-to-object and object-to-camera direction and distance in 3D), normalized score (0-100): IoU of predicted boxes or the L2 error of predicted points (pixels in 2D, metres in 3D) mapped to 0-100, on the simulated robotic tabletop scenes of Embodied3DBench (21K QA pairs over 4.8K object instances), zero-shot; higher is better. Source: arxiv.org. Saturation forecast: Around January 2027. 13 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | GPT-5 | 77.7 |
| 2 | Doubao-Seed-1.6 (Thinking) | 72.2 |
| 3 | Gemini 2.5 Flash | 67.4 |
| 4 | InternVL3-38B | 67.3 |
| 5 | Claude Sonnet 4 | 65.2 |
| 6 | Qwen 3 VL 4B Instruct | 65.2 |
| 7 | GPT-4o | 63.3 |
| 8 | Gemini 2.5 Pro | 60.2 |
| 9 | Qwen 2.5 VL 7B Instruct | 59.8 |
| 10 | Qwen 2.5 VL 72B Instruct | 56.5 |
Interactive version: theaggregate.ai/benchmark?slug=embodied3dbench-spatial-relation-3d · How It Works · Data refreshed daily, snapshot 2026-10-07.