Skip to main content
Open Images in the sidebar. The page has three columns: Settings on the left, the canvas with the prompt bar in the middle, and History on the right.
The AI Image Generator with presets, reference images, the prompt bar and the history panel

① Quick Presets · ② Reference Images · ③ Prompt bar · ④ Cost and generate · ⑤ History

​
The prompt bar

Everything starts in the prompt bar at the bottom of the canvas (③):
  • Prompt — describe the image: the subject, the setting, the style, the lighting and the framing. Two or three precise sentences usually beat a long paragraph.
  • Model — GPT Image 2.5 Flare by default. Click its name to open the model picker.
  • Enhance (the wand) — turns a short idea into a detailed prompt. You pick an enhancement style first.
  • Reference (the picture icon) — attach reference images without scrolling the settings.
  • Aspect ratio and number of images — the same options as in the settings panel, at hand.
  • Cost (④) — the credits this generation will use. Click the arrow to generate.
A good image prompt names who or what, where, the style and the light: “A ceramic coffee cup on a marble counter, morning window light, soft shadows, editorial product photo, shallow depth of field”.

​
Quick Presets

Quick Presets (①) are 21 professional templates — ready-made looks such as Viral Desk Figure or Fantasy Illustration. Click one to load its prompt and settings, then change the subject to yours. View all presets (or Browse all presets on the empty canvas) opens the full gallery.

​
Reference Images

Drop images in Reference Images (②) — or click Select from library — to guide the result:
  • Keep a character consistent across several pictures.
  • Show a product that must appear exactly as it is.
  • Match a style from an example image.
  • Edit an image — add it as a reference and describe the change: “make it night-time”, “put the logo on the cup”, “remove the person in the background”.

​
Settings

Image settings: quality, aspect ratio, number of images and the output summary

① Quality · ② Aspect ratio · ③ Number of images · ④ What you'll get · ⑤ Technical Settings

  • Quality (①) — Low, Medium (default) or High. Higher quality gives finer detail and cleaner text inside the image, and costs more.
  • Aspect ratio (②) — 9:16 for TikTok, Reels and Shorts, 1:1, 3:4, 4:3 or 16:9.
  • Number of images (③) — 1 to 4 variations in one go. The cost is per image.
  • What you’ll get (④) — the output resolution, the number of images and the total credits, before you spend anything.
  • Technical Settings (⑤) — extra options for the selected model. Use What to avoid to list what you don’t want in the picture (text, watermark, extra fingers).

​
Pricing

With the default model, each image costs: Other models have their own prices, shown on the button before you generate.

​
History

The History panel (⑤) lists every generation by date, with the model, the prompt, the time, the credits and the quality. Click a result to open it large and download it. See all generated images opens them in your media library.

​
Available models

The model picker includes GPT Image 2.5 Flare, GPT Image 2, GPT Image 1.5, Nano Banana (2.1, 2, Pro), FLUX.2 Pro, FLUX Pro Ultra, FLUX 2 Turbo, Imagen 4 Ultra and Fast, Seedream V5 Pro and Lite, Seedream 4.5, Ideogram V3, Kling 3.0 Image, Kling Omni 3, Grok Imagine Image, Hunyuan v3 Image and more. The list grows as new models launch.
Use Ideogram or GPT Image when you need readable text inside the image, and reference images when a character must look the same in several pictures. For story characters you’ll reuse in videos, create them in Characters.