ImgsGenAt a glance
GPT Image in ImgsGen offers draft, standard and fine quality; 1K, 2K and 4K targets; custom dimensions; 1–4 requested images; and reference editing. Actual output dimensions and availability depend on the selected model and request.
Available models
The current studio catalog includes GPT Image 2, GPT Image 2.5 Flare and GPT Image 2.5 Sunburst. You can reuse a written prompt across models to compare how each interprets a scene. Model names identify the available options and do not imply a partnership with the provider.
Control quality and size
Select a resolution tier and aspect ratio, or set custom width and height. Custom dimensions must be multiples of 16, each side must not exceed 4096 pixels, and the request must fit the area and aspect-ratio limits shown by the studio. Quality is a separate setting.
Edit with a reference picture
Attach a picture and describe the elements that should remain and the changes you want. You can also select an image from history and use it as a reference. Inspect the revised subject, text and edges rather than assuming every detail will be preserved.
Check the finished result
Image details show the actual pixel dimensions. If a model adjusts the output size or returns a different image count, the studio displays a notice and saves the returned results. The linked comparison includes one GPT Image 2 draft-quality sample, not a benchmark of every model and setting.
Bring the idea to your canvas.
Choose a supported model and composition. The studio uses an ImgsGen account and subscription credits.
Open image studio ↗