Gemini 3.1 Flash with image generation capabilities. High-efficiency image generation model with support for text rendering, reference images, search grounding, and thinking mode. The efficient counterpart to Gemini 3 Pro Image.
Settled by community votes across 1 shared challenge, with an AI judge weighing in on each.
Nano Banana 2
29.0
arena score
#1 of 44 in Text-to-Image
Best Text-to-Image right now
Top 2 in Image Editing
Skill signature
Not enough comparable category data
The chart appears once both models have ratings across at least three shared arena categories.
GPT Image 1
13.3
arena score
#19 of 23 in Image Editing
Vote tally
Where the votes landed
Nano Banana 2
0%
win rate
Ties
0%
GPT Image 1
0%
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
EditingEdit 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
Nano Banana 2
GPT Image 1
AI Judge Analysis
Nano Banana 2
- + Excellent preservation of the specific vitiligo patterns on the hands through the edit.
- + Included the glasses from Image 2 which were technically part of the outfit.
- + Maintained the sandy texture on the face consistently with the source.
- − The glasses frames are slightly warped and blend poorly with the face and hair.
- − The background lighting is noticeably different from the original base image.
GPT Image 1
- + High visual quality with very clean integration of the new clothing items.
- + Maintained the person's facial features and vitiligo patterns on the forehead accurately.
- + The scarf pattern and coat texture are rendered with high detail.
- − Missed the sunglasses from the source outfit image.
- − The hand visible on the right lacks the prominent vitiligo pattern seen in the source image.
Verdict: Nano Banana 2 is more successful at preserving the specific skin details of the subject, particularly the vitiligo on the hands, and includes the sunglasses from the reference outfit. GPT Image 1 offers a cleaner overall aesthetic but fails to maintain consistent skin markings in the edited areas and omits an accessory.
Next steps
Explore each model
OpenAI's previous image generation model that accepts both text and image inputs and produces image outputs