GPT Image 2 OpenAI P-Image Edit PrunaAI

Settled by community votes across 1 shared challenge, with an AI judge weighing in on each.

GPT Image 2

28.2 arena score

#3 of 44 in Text-to-Image

Top 3 in Text-to-Image

Skill signature

Not enough comparable category data

The chart appears once both models have ratings across at least three shared arena categories.

P-Image Edit

0.2 arena score

#24 of 23 in Image Editing

Vote tally

Where the votes landed

GPT Image 2

win rate

Ties

P-Image Edit

win rate

Shared challenges 1

Challenge by challenge

The strongest take from each model on every shared challenge, with the AI judge's read.

Outfit Transfer Challenge

Editing

Edit instruction

“Use Image 1 as the base person. Dress them in the exact elaborate outfit from Image 2 (including all layers, accessories, jewelry, and shoes). Carefully adapt the clothing to the body shape and pose in Image 1 while maintaining realistic fabric behavior, correct proportions, and perfect lighting/shadow matching. Keep the person’s exact face, hair, and background completely unchanged.”

Source

GPT Image 2

P-Image Edit

AI Judge Analysis

GPT Image 2

+ Excellent preservation of the original person's face, hair, and vitiligo patterns
+ High-fidelity recreation of the coat and scarf textures from Image 2
+ Natural lighting and shadows that integrate the person into the scene

− Cropped the person differently, losing the full-body view and shoes
− Subtle changes to the background details around the wooden structure

P-Image Edit

+ Successfully adapted the full-body pose including shoes from the prompt's implied scope
+ Kept the person's face and unique vitiligo markings intact

− Added chunky gold jewelry and yellow laces not present in Image 2
− The clothing colors are significantly oversaturated compared to the source image
− Visual glitches around the left eye/eyebrow and distorted finger anatomy

Verdict: GPT Image 2 (Model A) provides a much higher quality edit by perfectly matching the textures and style of the reference clothing while maintaining the person's identity and skin patterns flawlessly. While it chooses a more zoomed-in composition, it avoids the anatomical errors and color inaccuracies found in P-Image Edit (Model B), which added non-existent accessories and suffered from severe oversaturation.