PaintBench — leaderboard
Image-editing benchmark measuring pixel-precise object, background, and attribute edits with automated overall mean-IoU scoring.
Metric: Mean IoU (%). Source: paintbench.github.io. 11 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | Nano Banana 2 | 17.1 |
| 2 | GPT Image Edit 2 | 16.3 |
| 3 | Nano Banana 1 | 11.1 |
| 4 | Qwen-Image-Edit | 6.7 |
| 5 | BAGEL-7B-MoT | 5 |
| 6 | FLUX.2-dev | 4.6 |
| 7 | FLUX.1-Kontext | 3.9 |
| 8 | LongCat-Edit | 3.6 |
| 9 | FLUX.2-Klein-9B | 3.5 |
| 10 | HunyuanImage-3 | 0.4 |
Interactive version: theaggregate.ai/benchmark?slug=paintbench · How the rankings work · Data refreshed daily, snapshot 2026-07-22.