UniEditBench (Image) - Structural Fidelity: leaderboard
UniEditBench (2026) scores image edits on 633 samples across nine editing operations on a 1-5 scale with an evaluator the authors distilled from Qwen3-VL-235B-A22B into Qwen3-VL-8B. This board rates how well the edit keeps the source layout and identity, from 1 to 5; higher is better.
Metric: Structural fidelity score (1-5): the edit keeps the source image's layout and identity; 633 image-editing samples over nine operations, each edit scored on a 1-5 Likert scale by the authors' Qwen3-VL-8B evaluator distilled from Qwen3-VL-235B-A22B (the second entry of each 4B/8B cell); higher is better. Source: arxiv.org. Saturation forecast: Rough model projection: around 2026. 16 models tracked. Updated 7 Oct 2026.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | UniEditBench Hunyuan (checkpoint unspecified) | 4.69 | |
| 2 | FLUX.1-Kontext-dev | 4.63 | |
| 3 | DreamOmni2 | 4.56 | |
| 4 | Qwen-Image-Edit | 4.53 | |
| 5 | Step1X-Edit | 4.47 | |
| 6 | OmniGen2 | 4.38 | #4579 |
| 7 | UniWorld-V2 | 4.3 | |
| 8 | ICEdit | 4.25 | |
| 9 | BAGEL-7B-MoT | 4.17 | #4619 |
| 10 | LaVida-O | 3.89 | |
| 11 | VAREdit | 3.82 | |
| 12 | AnyEdit | 3.62 | |
| 13 | EditAR | 3.37 | |
| 14 | EditMGT | 2.61 | |
| 15 | InstructPix2Pix | 1.74 |
Interactive version: theaggregate.ai/benchmark?slug=unieditbench-image-structural-fidelity · How It Works · Data refreshed daily, snapshot 2026-11-21.