One prompt. 18 image models. Output test.
We gave every text-to-image model on QynTro the same prompt, a fashion photo of a woman in a hot-pink suit on a cobalt-blue car against a yellow wall, and asked each for a square picture. Here is every picture as it came back, cheapest first, with what it costs today.
- 18 models
- 1 take each, unedited
- From 1 credit a picture
- Made 4 Oct 2026
18 pictures, cheapest first
Click a picture to open it full size. Each price is what that picture costs on QynTro today, and Create opens the AI Image Generator with that model and this prompt.
GPT Image Mini
OpenAI- 1024 x 1024
- 1:1
2 credits for this picture
GPT Image 2.5 Flare
OpenAI- 1024 x 1024
- 1:1
2 credits for this picture
GPT Image 2.5 Sunburst
OpenAI- 1024 x 1024
- 1:1
2 credits for this picture
FLUX.3 Image
Black Forest Labs- 1024 x 1024
- 1:1
4 credits for this picture
Nano Banana 2 Lite
Google- 1024 x 1024
- 1:1
4 credits for this picture
Nano Banana Lite
Google- 1024 x 1024
- 1:1
4 credits for this picture
Nano Banana
Google- 1024 x 1024
- 1:1
4 credits for this picture
Grok Imagine 2.0
xAI- 1024 x 1024
- 1:1
5 credits for this picture
Nano Banana 2
Google- 1024 x 1024
- 1:1
7 credits for this picture
GPT Image 1.5
OpenAI- 1024 x 1024
- 1:1
8 credits for this picture
GPT Image 2
OpenAI- 1024 x 1024
- 1:1
8 credits for this picture
Nano Banana Pro
Google- 1024 x 1024
- 1:1
12 credits for this picture
Every picture is shown as a 1024-pixel WebP copy of the file the model made, so the page loads quickly. Click one to open it.
Same prompt, same settings, one take
One prompt, one picture
Every model got the same 323-character prompt and made one square picture. Nothing was picked from several tries or edited.
The same shape
A square from every model: 1024 x 1024 for GPT Image, and Square HD for every other model. Seedream returns its square at 2048 x 2048; this page shows every picture at 1024 pixels.
Today's prices
Each price is what the AI Image Generator charges today for one picture, read from the tool itself. The tool shows the price before you make anything, and a picture that fails is not charged.
Questions about this test
Which model made the cheapest picture?
Were the pictures edited, or picked from several tries?
Can I make the same picture myself?
Why are two of the pictures bigger?
Your prompt, any of these models
One QynTro account runs every model on this page, plus the rest of the image, video, music and voice tools.