InEdit-Bench - State Transition: leaderboard
Metric: Average (0-100) of the four dimensions (appearance consistency, perceptual quality, semantic consistency, logical coherence) on InEdit-Bench's state transition tasks, where the editing model generates the intermediate stages of a multi-step transformation from an initial image to a described final state, scored 0-100 by a GPT-4o-2024-11-20 evaluator; higher is better. Source: arxiv.org. Saturation forecast: Rough model projection: around 2026. 14 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | GPT-Image-1 | 81.12 | |
| 2 | Gemini 2.5 Flash Image (Preview) (Nano Banana) | 70.15 | |
| 3 | Flux-Kontext-pro | 51.76 | |
| 4 | Qwen-Image-Edit | 44.77 | |
| 5 | BAGEL-7B-MoT (Thinking) | 41.84 | |
| 6 | OmniGen2 | 38.39 | |
| 7 | SeedEdit 3.0 | 38.36 | |
| 8 | BAGEL-7B-MoT | 38.27 | |
| 9 | Emu2 (InEdit-Bench checkpoint unspecified) | 33.46 | |
| 10 | InstructPix2Pix | 28.57 |
Interactive version: theaggregate.ai/benchmark?slug=inedit-bench-state-transition · How It Works · Data refreshed daily, snapshot 2026-10-11.