GPT Image in Doppelshot
GPT Image 2.5 is OpenAI's image model, the one behind images in ChatGPT. Doppelshot runs its two flavours, Flare and Sunburst, as photo engines: detailed scenes, painterly and vintage styles, and text that comes out readable.
What it is
- Follows a long, structured prompt well: scene, outfit, light, mood, constraints, in that order.
- Good at styles that are not photos: oil on canvas, film stills, renaissance portraits, vintage prints.
- Writes short text in the picture cleanly, which helps posters and covers.
What Doppelshot makes with it
- The default engine of the renaissance portrait and vintage film cards.
- An alternative engine on every photo card: switch in the Studio when the look wants painting or film.
- Twin photos through Phota: your learned face profile rendered by GPT Image.
Engines and prices
| Engine | Kind | Price | Cards |
|---|---|---|---|
| Live Edit GPT Flare | Photo | 1 credit | Studio headshot, Just a picture, Try on +18 |
| Live Edit GPT Sunburst | Photo | 1 credit | Studio headshot, Just a picture, Try on +18 |
| Twin Photo (Phota, GPT Flare) | Twin photo | 3 credits | Just a picture, Studio headshot, Try on +18 |
| Twin Photo (Phota, GPT Sunburst) | Twin photo | 3 credits | Just a picture, Studio headshot, Try on +18 |
Checked against the catalog on 2026-10-21.
Cards that run on it
Open a card, drop your selfie, and pick this engine in the Studio when it is not the default.
1 credit
1 credit
top1 free
new1 free
3 credits
2 credits
1 free
2 creditsQuestions
When should I pick GPT Image over Nano Banana?
For a style (painting, film, illustration) or a busy scene with many elements. For a plain realistic photo, Nano Banana is faster and just as faithful.
Is my selfie sent to OpenAI?
Only to produce the generation, through a short-lived signed link, and never to train. The result stays in your private library.
Try it on your selfieThe first preview is free, no card needed.
Other models
from 1 creditsNano Bananaby GoogleGoogle's image models: fast, faithful to your face, up to 4K with Pro.from 1 creditsIdeogramby IdeogramTypography that reads: posters, covers and any image with exact text. Twin training too.from 2 creditsSeedreamby ByteDanceByteDance's image model: large, detailed, strong on fashion and products.from 1 creditsKreaby KreaEdits that look like a real camera roll, no plastic skin.