GPT Image 2.0AI Image Generator

Create and edit images from a prompt or reference image in the RSW AI Studio workspace.

Workspace

1 Credits
Model
No file selected

JPG, PNG, WebP. Up to 16 images, max 10MB each.

No reference image uploaded yet.

Try 365 DAYS Unlimited Seedance 2

Inspirations

Editorial Portrait

Editorial Portrait

A cinematic portrait in soft natural light, 85mm lens look, subtle film grain, editorial style.

Product Poster

Product Poster

A premium product poster with dramatic spotlight, clean dark gradient background, high contrast.

History

No history yet. Generate your first task.

One photo. A new point of view.

Slide to explore a look. Choose Create similar, then add your photo above.

Illustrative comparisons made in ChatGPT. Your results will vary.

  • The street reimagined with carved outlines and textured woodblock colors
    Photographic Japanese shopping street with a yellow van and pedestrians

    Woodblock streets

  • A recognizable miniature figure in original colorful toy packaging
    Fictional adult wearing a teal overshirt and cream trousers

    Your collectible edition

  • The same portrait interpreted with expressive anime linework and cel shading
    Fictional adult with short dark hair in a rust jacket under spring trees

    An anime moment

  • The village in a luminous hand-painted storybook interpretation
    A winding path through a quiet coastal hillside village in daylight

    A storybook escape

  • The same dog drawn with soft wax crayon strokes on textured paper
    A small tan and white dog sitting on a rug beside a window

    Crayon companion

  • The same person with soft studio lighting and understated business clothing
    Natural casual portrait of a fictional adult in a simple dark shirt

    A fresh studio portrait

  • A bold graphic portrait with halftone shading and colorful shapes
    Fictional adult with wavy dark hair in a saffron blouse

    Pop-art energy

  • The same family as friendly original three-dimensional animated characters
    A fictional family of three smiling together in a sunny garden

    Family, reimagined

What is GPT Image 2.0?

Turn a clear brief into an image, or give an existing photo a new direction.

GPT Image 2.0 is the name used in this workspace for GPT Image 2, OpenAI’s image generation and editing model. The official model overview describes support for text and image inputs. Here, you can work with a written prompt alone or add reference images to guide the result.

Use the same focused workspace for a product scene, a poster concept or a room restyle. Choose your framing and resolution, describe what matters, then inspect the image in History. For edits, say both what should change and what should stay—especially labels, faces and the position of the subject.

Reference images per request
16
Image per generation
1
Resolution presets, with supported ratios
1K–4K
Credits for a 1K image
1

Start with an idea. Or start with an image.

Two ways to create with GPT Image 2.0, without switching tools.

Create from text

Build a scene from a brief

Describe the subject, setting, composition and light. Specify exact wording for a poster, or leave room for your own typography. Start with one clear visual idea before adding more details.

  • Product concepts
  • Posters
  • Editorial illustrations

Edit with references

Change the scene, keep the essentials

Upload a photo you have permission to use. Ask for a new setting, color palette or material, and name the details to preserve. With several references, explain which image provides the subject and which provides the style.

  • Background changes
  • Room restyles
  • Material studies

Available in this workspace

References
JPG, PNG or WebP, up to 10MB each
Framing
Fixed aspect ratios, including square at 1K and 2K
Output
One image at 1K, 2K or 4K
Review
Preview and download completed images from History

How to use GPT Image 2.0

Start in the workspace above. A good first image begins with a focused brief.

  1. Describe the result

    Name the subject, its setting and the visual style. For an edit, add your reference and explicitly list the details to keep.

  2. Choose the canvas

    Match the aspect ratio to where you will use the image. Choose 1K, 2K or 4K, then review the credit quote on Generate Image before submitting.

  3. Review and refine

    Open the completed result in History. Check spelling, edges and any details you asked to preserve. Adjust one part of the prompt at a time, or download the image when it works for you.

GPT Image 2.0 prompt examples

Four briefs to adapt. Copy a prompt, add any required photo, and make it your own.

A clean product photograph

Text only · 1:1 · 1K or 2K

Photograph a single cobalt-blue ceramic espresso cup on a warm off-white tabletop. Use a low three-quarter camera angle, soft window light from the left and a gentle contact shadow. Place the cup slightly right of center with generous empty space on the left. Show realistic ceramic texture. No lettering, logos, extra cups or decorative props.

A coffee shop opening poster

Text only · 3:4 · 1K or 2K

Design a vertical opening-day poster for a fictional neighborhood coffee shop. Use warm cream paper, dark espresso lettering and a single terracotta cup illustration. Exact headline: "GOOD MORNINGS START HERE". Beneath it, place "OPENING SATURDAY" in smaller type. Use a clear reading order, generous margins and restrained print texture. Add no other words, prices or logos.

A room with a warmer palette

Upload a room photo · Match its aspect ratio

Restyle the room in my uploaded photo with warm oak, natural linen and muted olive accents. Keep the original camera angle, wall positions, windows, doors and main furniture placement unchanged. Replace only the visible finishes and soft furnishings. Use believable daylight and shadows consistent with the windows. Do not add rooms, people, text or extra windows.

A paper-cut portrait

Upload a portrait you have permission to use

Reinterpret the person in my uploaded portrait as a layered paper-cut illustration. Keep the face shape, hairstyle, expression, clothing colors and shoulder framing recognizable. Use carefully cut matte paper in coral, navy and cream, with subtle shadows between layers and a plain background. Avoid photographic skin texture. Add no text, accessories or additional people.

GPT Image 2.0 credits

Standard credit quotes for one GPT Image 2.0 image, by resolution.

Standard credit quotes for one GPT Image 2.0 image, by resolution.
Output1K2K4K
1 image1Default23

Choose the size you need

Use 1K for an initial concept. Select 2K or 4K when you need a larger output; a higher resolution does not guarantee that every detail will be correct.

Check the final quote

The Generate Image button reflects your settings and account eligibility. These figures describe standard credit pricing, not a change to your membership benefits.

Check the quote for your settings

Frequently asked questions

Are GPT Image 2 and GPT Image 2.0 the same model here?

Yes. This page labels the model GPT Image 2.0; its model identifier is gpt-image-2. GPT Image 2.5 has a separate page and different controls.

Can I edit a photo instead of generating from scratch?

Yes. Add the photo under Reference Images and describe the edit. Be specific about what to preserve. Reference guidance is not a pixel-perfect lock, so review faces, product details and lettering before using the result.

Why are some aspect ratios unavailable at 4K?

The current GPT Image 2.0 workspace supports 16:9, 9:16, 2:1, 1:2, 21:9 and 9:21 at 4K. For a square or 3:4 image, use 1K or 2K. The available ratio options follow your resolution selection.

Can I combine several reference images?

You can attach up to 16 JPG, PNG or WebP images, each up to 10MB. Explain each reference’s role rather than asking the model to guess—for example, the object from the first and the setting from the second.

How can I improve text in a generated poster?

Keep the wording short, put the exact text in quotation marks and explain its position and hierarchy. Check every letter afterward. For critical small print, leave a clear area and add that text in your design editor.

Do I need a ChatGPT plan or my own API key?

No. RSW AI Studio is a separate service with its own access and billing. You do not need to connect a ChatGPT subscription or provide an OpenAI API key to use this workspace.

Should I use GPT Image 2.0 or GPT Image 2.5?

This workspace offers a focused prompt, reference, ratio and resolution workflow. GPT Image 2.5 adds Flare and Sunburst choices plus additional quality and output controls. Compare the same brief on both if those controls matter to your work; their settings and prices are separate.

Compare results on another model, or go back to GPT Image 2.0.

Planning a retro portrait? Explore our 1980s photo prompts and customization guide for reference-photo tips and focused editing briefs.