HOI-Edit - Foundational Editing: leaderboard
HOI-Edit (2026) asks image editing models to change human-object interactions, such as picking up, using or moving an object, and scores the edit with grounded questions to Gemini 2.5 Pro about tracked human and object regions. This board is the 0 to 100 interaction score on 357 edits that create, remove or change one interaction; higher is better.
Metric: Interaction score (0-100) on the 357 Level-1 edits (create, remove or change one human-object interaction), scored by HOI-Eval: the human and object boxes are tracked into the edited image and Gemini 2.5 Pro rates the interaction from tagged crops; higher is better. Source: arxiv.org. Saturation forecast: Rough model projection: around 2026. 8 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Nano Banana | 72.71 | |
| 2 | Wan2.2-I2V-A14B | 69.08 | |
| 3 | Bagel | 63.26 | |
| 4 | Qwen-Image-Edit PLUS | 61.28 | |
| 5 | ChronoEdit | 58.23 | |
| 6 | Step1X-Edit | 53.96 | |
| 7 | FLUX.1 Kontext [dev] | 49.61 | |
| 8 | ByteMorph | 44.69 |
Interactive version: theaggregate.ai/benchmark?slug=hoi-edit-foundational-editing · How It Works · Data refreshed daily, snapshot 2026-11-13.