PhysEditBench - Depth: leaderboard
Metric: AbsRel-AI: absolute relative depth error after polarity selection and affine alignment, source-balanced over OpenRooms-FF (1,125 images) and InteriorVerse (2,109); prompted image editors on the paper main-axis split (indoor OpenRooms-FF and InteriorVerse scenes; official access settings of Appendix Table 19); task-specific specialist models are reference rows, not published; lower is better. Source: arxiv.org. Saturation forecast: Rough model projection: around 2026. 4 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | wan2.7-image | 0.17 |
| 2 | GPT-Image-2 | 0.17 |
| 3 | doubao-seedream-5-0-260128 | 0.17 |
| 4 | GPT-Image-1.5 | 0.23 |
Interactive version: theaggregate.ai/benchmark?slug=physeditbench-depth · How It Works · Data refreshed daily, snapshot 2026-10-07.