The chat can generate images as well as write. Ask for an image of something and the agent generates it, shows it in the conversation, and puts it in a note when you ask.
Asking for one
Ask the way you would ask a person: generate an image of a taco, make an illustration of a lighthouse at dusk, give me a poster for the study group. The chat shows a generating step while the image is made, and the image appears under it as soon as it is done. Describe the subject, the setting and the mood if you have one in mind, and say so if the image should hold no words.
Putting it in a note
Say where it goes: add that to the note, put the image under the ingredients, make a new note about tacos with the image in it. The agent places it the way it places any other edit, section by section, and the note keeps the image the way it keeps a photo you dropped in: it syncs with the note, and the note shows it on every device.
Which model generates it
Open Settings, go to Models and switch on one model under Image Models. Grape Pro Image Model is included with the Pro plan and needs no key. With your own API key you can pick OpenAI, Google or xAI instead, on Pro or Lifetime, and add a model that is not listed with Add Custom Model. The same model draws the pictures in a slide deck and the infographics.
The allowance
Images generated by Grape Pro Image Model count against the AI images bar in the usage dialog. Images generated with your own key are billed by that provider and are not metered here.
When it will not generate
If no image model is switched on, the agent says so and points you at Settings. If the provider refuses an image, rephrase it; the agent will not retry the same wording on its own.