Image model
Generate or transform one GPT Image with verified controls
GPT Image is a formal image model in WAN3. Start from text or attach exactly one source image, choose a supported aspect ratio, review the fixed credit price and submit from the same private Workspace.
01
Use text-to-image or image-to-image
Text-to-image accepts a prompt without media. Image-to-image requires exactly one JPEG, PNG or WebP source image up to 30 MB; the source remains attached to the private task record.
02
Choose from the verified aspect-ratio set
GPT Image supports Auto and a broad set of square, portrait, landscape and panoramic aspect ratios. The current verified contract creates one image and does not expose an unverified resolution selector.
03
Pay 3 credits for each submitted task
Text-to-image and image-to-image each cost 3 credits. A terminal provider failure refunds the task once; a successful result is copied to WAN3-managed storage before owner-only download.
04
Keep each result in private Creations history
Queued, processing, succeeded, failed and refunded status remains visible with the model, prompt and actual credit fact. Image records appear alongside video without making source material public.