OpenAI · Image model
GPT Image 2 AI image generator
OpenAI's image model. It follows a long written brief closely and puts readable text in the picture, with detail up to 4K.
Runs in Generate, in a workflow step and in batch · The cost shows before you press Generate · No separate OpenAI subscription
- Claude
- ChatGPT
- Cursor
At a glance
GPT Image 2 in DesignerBox
Use GPT Image 2 in Generate, in a step of your own workflow, or on many rows at once in batch.
- Made by
- OpenAI
- Type
- Image model
- You give it
- A text prompt, and up to 5 reference pictures
- Shapes
- 1:1, 4:3, 3:4, 16:9, 9:16
- Sizes
- Up to 4K
Strengths
What GPT Image 2 is good at
- Follows long briefsGive it a brief with many details and it keeps most of them. Good for layouts with several parts.Try it in Generate
- Text in the pictureHeadlines, prices and labels come out readable, so it suits ads with words.Try it in Generate
- Detail up to 4KLarge enough for a print or a big banner.Try it in Generate
- Photoreal picturesScenes and products look like a photo.Try it in Generate
How to use it
How to use GPT Image 2 in DesignerBox
- 1. Open Generate. Sign in to DesignerBox, open Generate and choose Picture. You can also add an image step to a workflow.
- 2. Choose GPT Image 2. Pick GPT Image 2 in the model list. Then set the shape and the number of pictures.
- 3. Write the brief and run. Describe the picture and add up to 5 reference pictures. The cost shows next to the button before you press Generate.
Start free. You see the price before you make anything.
Your own photos, your own prompts and the commercial licence start on the Pro plan.
Tips
Tips for better results with GPT Image 2
- Write the brief as a list. One line for the product, one for the background, one for each line of text.
- Check skin and fabric. Close-ups of skin and cloth are its weaker side. Try Seedream 5 for on-model shots.
- Give the text its place. Say where the words go: "headline at the top, price bottom right".
- Start from your own product photo. Add the real product as a reference. The model keeps its shape, label and colour closer than a description can.
- Name the shot. Say where the product stands, what is behind it and where the light comes from. Words like "premium" give the model nothing to draw.
- Save the brief as a workflow. When one picture works, save the steps as a workflow. Every next product then runs the same brief on the same model.
Other models
Other image models in DesignerBox
Every model below runs in the same workspace, on the same plan. Also built in: Kontext Multi.
Nano Banana ProGoogle's image model for fine detail, readable text and editsGoogle
Nano Banana 2Google's fast image model for drafts and large setsGoogle
Seedream 5ByteDance's image model for people, clothing and text, at 2K to 3KByteDance
FLUX 2 FlexTypography, retexturing and edits from several referencesBlack Forest Labs
FLUX Pro 1.1Steady, photoreal pictures for large setsBlack Forest Labs
What is GPT Image 2?
GPT Image 2 is an image model from OpenAI. It follows long written briefs and renders readable text, with detail up to 4K.
Can I use GPT Image 2 in DesignerBox?
Yes. Pick it in Generate or in an image step of a workflow.
Which shapes does GPT Image 2 make?
Five: 1:1, 4:3, 3:4, 16:9 and 9:16.
When should I pick another model?
For people and clothing, Seedream 5 is the better choice. For the most careful detail on a product, use Nano Banana Pro.
Make your first pictures today.
Sign up with Google or your email. Nothing to set up.
Get started free- Claude
- ChatGPT
- Cursor
















