RE-Edit - Cultural: leaderboard
RE-Edit (2026) tests instruction-based image editors on 1,000 curated requests whose correct edit depends on implicit reasoning. This board is the pass rate on the cultural-reasoning subset (cultural conventions), with a Qwen3-VL-30B judge marking each edited image pass or fail against the case rationale. Higher is better.
Metric: Pass rate (%) on the cultural-reasoning subset of RE-Edit's 1,000 curated edit requests (cultural conventions): a Qwen3-VL-30B judge marks each edited image pass or fail against the case rationale; higher is better. Source: arxiv.org. Saturation forecast: Rough model projection: around 2026. 12 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | FLUX.2-dev | 7.8 | |
| 2 | Seedream 4.0 | 6.1 | |
| 3 | Nano Banana | 3.9 | |
| 4 | Qwen-Image-Edit | 3.3 | |
| 5 | FLUX.1 Kontext [dev] | 2.8 | |
| 6 | Step1X-Edit-v1p1 | 2.2 | |
| 7 | DreamOmni2 | 2.2 | |
| 8 | Ovis-U1-3B | 2.2 | #2318 |
| 9 | OmniGen2 | 1.1 | #4321 |
| 10 | HiDream-E1 | 1.1 |
Interactive version: theaggregate.ai/benchmark?slug=re-edit-cultural · How It Works · Data refreshed daily, snapshot 2026-11-13.