Current-product comparison · checked September 26, 2026
OpenAI has retired the official DALL·E GPT from ChatGPT. Its current image experience is ChatGPT Images, where creation and editing happen through conversation, image upload, optional area selection, and repeated instructions. Stable Diffusion AI is different: it gives you separate generator, Image to Image, Photo Editor, and background tools instead of a conversational workspace.
Try this workflow
Write the result you need, choose a ratio, and run it below. You can revise the prompt, retry, download, transform the result, or open it in Photo Editor. This is the current product experience; it is not a DALL·E endpoint and does not promise the same output style or instruction following.
Image preview
ExamplePrompt inspiration
A small wooden cabin in fresh snow at blue hour, warm light through the windows, pine forest, wide cinematic composition
Choose by task
ChatGPT Images
Describe the image in conversation; follow-up questions and later messages can refine the direction.
Stable Diffusion AI
Write one visible prompt, choose an aspect ratio, generate, review, and revise the prompt directly.
ChatGPT Images
Upload an image, describe changes, and optionally select an area before continuing the conversation.
Stable Diffusion AI
Use Image to Image for a broader transformation or Photo Editor for a focused instruction with before-and-after review.
ChatGPT Images
Current documentation highlights adding text and offers image templates in supported contexts.
Stable Diffusion AI
Use the generator for visual concepts, but plan to verify or typeset important words separately.
ChatGPT Images
Keep editing through the same conversation and save from the image experience.
Stable Diffusion AI
Download, transform, edit, save eligible results in Creations, or move into a purpose-built background workflow.
Its current help documentation describes generating in a conversation, uploading an image, asking for changes, selecting part of an image, adjusting aspect ratio, and continuing to revise through chat. That makes it the natural choice when the back-and-forth reasoning and remembered conversation are part of the job.
Availability and tier-specific options can change; check OpenAI’s current help page before planning a production workflow.
The generator keeps the prompt and ratio visible. Image to Image separates a source image from the transformation direction. Photo Editor separates what to change, where it is, and what to preserve. Purpose-built background tools add transparent PNG, white canvas, product layout, headshot, car, or bulk outcomes.
Honest limits
ChatGPT Images is documented as able to add text and follow template-like instructions. This site can generate poster or logo concepts, but generated lettering, labels, and exact layout still require close review and may need a design tool afterward.
ChatGPT offers an area-selection editor. Photo Editor here uses written location and preservation instructions without a pixel mask, so a change can extend beyond the intended area.
Official source
OpenAI’s help page is the source for the present ChatGPT Images experience and explicitly notes that the official DALL·E GPT has been retired. This guide deliberately avoids stale DALL·E 3 pricing, access, or performance claims.
Images in ChatGPT ↗