VMetaphor-Bench - Perceptual Harmony: leaderboard

Metric: Judge rating (1-5) (Qwen3.5-27B rates each generated image from 1 to 5 against the metaphor's structured annotation, 1,500 visual metaphors; this board is the dimension perceptual harmony: how naturally and coherently the two domains are blended and composed; conceptual prompts that state only the metaphorical idea). Source: arxiv.org. Saturation forecast: Rough model projection: around 2026. 11 models tracked.

Top models

#ModelScore
1GPT Image 1.54.38
2Nano Banana 2 (Gemini 3.1 Flash Image Preview)4.21
3Seedream 5.0 Lite4.08
4FLUX.2-dev3.93
5Qwen-Image-25123.84
6FLUX.1-dev3.82
7Stable Diffusion 3.5 Large3.75
8Z-Image-Turbo3.67
9BAGEL-7B-MoT3.56
10OmniGen23.45

Interactive version: theaggregate.ai/benchmark?slug=vmetaphor-bench-perceptual-harmony · How It Works · Data refreshed daily, snapshot 2026-09-26.