sangtao.ai
API key

Images

Making images with GPT Image

Write a prompt, pick the number of images, aspect ratio and resolution, then create with GPT Image 2.6 on the image page.

The image page turns a text description (a prompt) into an AI image with GPT Image 2.6. The prompt box sits at the top; below it is a gallery of community samples to borrow ideas from. Anyone can look around and try the controls — you are asked to sign in only when you press create.

The /imagine page: prompt box with a prompt typed, the chip row below it (paperclip, GPT Image 2.6, image count, ratio, Standard/2K) and the round arrow create button

Creating an image

Step 1. Describe the image

Type into "Describe the image you want…". The more specific you are (subject, setting, style, colours, any text that should appear), the closer the result. The box grows with your prompt; Enter starts a new line.

Step 2. Set the chips

  • Images (picture icon): how many images one press makes. Paid plans can pick up to 8 per press; free mode makes a fixed number.
  • Aspect ratio: pick from the Portrait / Landscape columns or square, or switch on Auto to let the model choose. The preview shows the frame's shape before you create.
  • Resolution: Standard (the model's native size) or 2K (long edge about 2,048px). 2K needs a paid plan.
  • Paperclip: add reference images — see Reference images.

Step 3. Choose visibility

The icon beside the create button shows whether the image will be Public (shown on the community page — anyone can see and reuse it) or Private (only you can see it). Private needs a paid plan.

Step 4. Create

Press the arrow button (or Ctrl+Enter). The label next to it always says what the press costs: Free, what is left on your plan, or a number of credits. Images in progress appear in the image tray at the bottom right, with your place in the queue and an estimated time — the tray follows you to other pages.

When the image is ready

In the tray each image has View, Download, Edit (use the image as a reference and keep editing) and Remove background. A failed image has Retry. Everything you made is also in History.

Creating for free

Besides credits and monthly plans, the image page has a free mode. What the screen tells you:

  • Images made for free are shown publicly on the community page.
  • The first time, you verify your email (a 6-digit code sent to you) and agree to the Free image terms — each only once.
  • One free run at a time: wait for it to finish before starting another. Free images run in their own queue; when there is a wait, the screen shows the time until your next image.
  • Free mode takes up to 3 reference images, and has no 2K and no Private.
  • A free image that fails doesn't count against you.

Cost

Each image comes off your monthly plan if you have one, otherwise off your credits — currently 3 credits per image. When your plan has fewer images left than you picked, the label beside the button says how many come off the plan and how many cost credits. Current prices and plans: see Pricing.

Under the prompt box is a gallery sorted by topic (Art, Product, T-shirt & POD, Logo…) with a Search samples… box. Open an image to see its prompt, press Use this sample to load the prompt and ratio into the prompt box, or Copy prompt. Tap ♥ to save an image to Favorites.

A sample opened from the gallery: its prompt and the "Use this sample", "Edit", "Remove BG" and "Download" buttons

See also