The new ChatGPT image model turns prompts, sketches, and uploads into production-ready visuals. Ask in plain language, refine with follow-ups, and keep people, products, and text sharp.
Examples: See what Nano Banana can do1 / 9
Before

After

Prompt used:
"Dismantle the wardrobe into its constituent parts."
Before

After

Prompt used:
"Remove the object from the image."
Before

After

Prompt used:
"Change the bikini to red"
Before

After

Prompt used:
"Transform this anime character into a collectible figure product showcase: Create a physical PVC figure standing on a clear round base, place a product box with the character artwork behind it, and add a computer monitor showing the 3D modeling process in Blender."
Before

After

Prompt used:
"Repair and color this photo"
Before

After

Prompt used:
"Transform the subject into a handmade crocheted yarn doll with a cute, chibi-style appearance."
Before

After

Prompt used:
"Reimagine the photo in the style of Van Gogh's 'Starry Night'."
Before

After

Prompt used:
"Change the hair to blue."
Before

After

Prompt used:
"Transform the person into a LEGO minifigure, inside its packaging box."
Swipe or use arrows to see more examples
GPT Image combines OpenAI’s latest image model with ChatGPT so you can ideate, edit, and publish marketing visuals from one place.
Add, remove, and restyle elements with layout-aware instructions that preserve composition.
Use the canvas, rough drafts, or product shots; ChatGPT keeps proportions and scene structure intact.
Improved fidelity for faces, hands, signage, and packaging so assets look realistic.
Generate up to 4 options at once, compare, and pick the render that fits your campaign.
Move from idea to downloadable assets in four quick steps.
Pick the GPT Image model and choose your aspect ratio.
Type the scene you need or upload a sketch/product photo to guide composition.
Ask for lighting, wardrobe, copy changes, or retakes until it matches the brief.
Export the selected render, save prompts, and reuse references for the next shoot.
Designed for teams that need realistic, editable visuals without leaving ChatGPT.
Understands spatial instructions—subjects stay framed, lighting stays natural.
Better detail for faces, hands, hair, and fabric for lifelike shots.
Generates signs, packaging, and captions that match your copy.
Upload moodboards, diagrams, or product shots and keep the layout intact.
OpenAI safety filters plus on-image provenance signals for responsible use.
Download high-res renders or plug the model into automated flows.
Answers to the top questions about OpenAI’s new image model.
Ship campaign-ready visuals with prompt, upload, and sketch-based control in ChatGPT.