AIM-Bench (Affective Image Manipulation) - Negative Emotion Accuracy: leaderboard

AIM-Bench emotion accuracy on edits whose target is a negative emotion (disgust, sadness, fear or anger), judged by Gemini-2.5-Pro emotion classification. Percent, higher is better.

Metric: Emotion accuracy (%) on the items whose target is a negative emotion (disgust, sadness, fear, anger): Gemini-2.5-Pro classifies each edited image into the 8 Mikels emotions, over the 800 AIM-Bench affective image editing items (EmoSet source images, a target Mikels emotion with target valence-arousal-dominance coordinates and an editing instruction; 8 emotion categories, 5 editing types), each model run once with its official default settings; higher is better. Source: arxiv.org. Saturation forecast: Rough model projection: around 2026. 13 models tracked. Updated 7 Oct 2026.

Top models

#ModelScoreOverall rank
1Step1X-Edit (AIM-Bench checkpoint unspecified)70.1
2UniWorld-V258.1
3Seedream 4.056.74
4BAGEL-7B-MoT56.61#4668
5FLUX-Kontext-Max55
6Qwen-Image-Edit-Plus (AIM-Bench checkpoint unspecified)53.75
7Qwen-Image-Edit-250950.76
8SeedEdit 3.050.53
9FLUX.1-Kontext-dev49.13
10FLUX-Kontext-Pro47.71
11OmniGen247.13#4625
12DreamOmni231.42
13InstructPix2Pix18.95

Interactive version: theaggregate.ai/benchmark?slug=aim-bench-affective-image-manipulation-negative-emotion-accuracy · How It Works · Data refreshed daily, snapshot 2026-11-22.