Images
Create images in a conversation, change them with follow-up prompts, and pull them straight into articles — all from one library.
What you can do
Images turns a written description into a finished image, as a chat. Describe what you want, pick an AI model, and the image comes back as a reply. To change it, send a follow-up prompt — the image becomes the starting point for an edit and a new image is created, so the original is always kept. Keep going for as long as you like; every step is its own image. Each one is saved to a personal library where tagging, search, and article integration are built in.
Supported AI providers
Image generation works with any connected provider that offers image-capable models — OpenAI, Google (Imagen + Gemini image models), and xAI (Grok Imagine). The picker shows the current model lineup for each vendor.
Creating an image
- Select a model. Only providers with image-capable models are shown.
- Configure settings. Available options depend on the model — size, aspect ratio, quality level, or resolution.
- Write a prompt and send it. Be specific about style, composition, colors, and subject — type it or dictate it by voice. The image comes back as a reply in the conversation.
New prompts are automatically enriched with your style settings before they reach the model — that's the Enhance prompt switch in the + menu. Turn it off to send your exact wording untouched.
Editing an image
To change an image, send a follow-up prompt describing what you want — e.g. "make the sky more dramatic" or "add a boat in the foreground" — or use the Edit action on any saved image. The image becomes the starting point and a new image is created, so the original is never overwritten. Keep sending prompts to iterate as many times as you like; each step is its own image in the library, linked to the same conversation so you can return to it later.
You can also start from a picture of your own instead of a prompt: attach an image with the + button or paste it straight in, describe the change, and the model transforms it — restyle a photo, swap a background, or refresh a product shot. The model picker automatically narrows to the providers that support image editing.
Models without direct editing (such as Imagen) still follow the conversation — instead of editing the image, they recreate it from everything described so far, so the result matches the description rather than the exact original. The label above the input shows which you'll get before you send: Editing keeps the same image and changes it; Reimagining creates a fresh one from the accumulated description.
Tagging and organizing
Each generated image can carry one or more tags. Tags appear as filter pills on the My Images view and let you scope the library to a project, a brand, or a campaign. Tag management lives in the same view — rename a tag and every image carrying it updates; delete a tag and it's removed from every image (the images themselves stay).
Using generated images in articles
In the article editor, the image picker includes a My Images tab alongside Generate, Search, and Upload — pick any saved generation as the article's featured image without leaving the page.
Usage
Each image you create — including every edit — counts as one use against the daily image limit on your plan. See Plans & usage limits.