WorldEdit: leaderboard
Metric: Overall average (1-5) on WorldEdit-Test, implicit causal image-editing instructions across ten transformation types and an other category, the equal-weight mean of the eleven category averages: mean of four 1-5 scores from a Qwen-VL-Max judge (KRIS-Bench protocol): visual consistency with the input image, visual quality, instruction following and knowledge plausibility; higher is better. Source: arxiv.org. Saturation forecast: Around January 2027. 12 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | GPT-4o | 4.36 | #333 |
No result here: #3 Claude Opus 5.5, #5 GPT-6 Astra, #8 Claude Fable 5.1.
Interactive version: theaggregate.ai/benchmark?slug=worldedit · How It Works · Data refreshed daily, snapshot 2026-10-11.