To content

← All guides

AI companion pictures: how image generation works and what it costs

Cleo · editorial persona · November 16, 2026

How image generation works in companion apps

The basic flow is simple. You write a short description or pick a few settings, and a model draws a picture from those words. The picture is new each time, so two requests with the same text can give two different results. Retries are part of the process.

Here is one sequence to compare with the provider you use:

  1. You send a text request, with optional settings such as pose, outfit or setting.
  2. The provider may check the request against its moderation rules. Candy.ai’s Terms say moderation controls may moderate or alter a request before an output is generated.
  3. A model renders the image.
  4. You see the result. It can be imperfect: hands, lighting or small details may not match what you asked for, so you may regenerate.

Read this sequence on the provider’s own site rather than assuming it, and check whether a retry works the same way as the first attempt.

What can cost extra

Image generation may be an add-on, or it may be included in a plan. Pricing differs by provider, so check these points on the provider’s pricing page:

Candy.ai’s Terms say subscription tokens can be used for extended features such as image generation, and that pre-made media can also be purchased. These are two separate paths: spending tokens you already have, or buying a finished item. Read both before you add anything to a cart.

Do not try to produce images that look like a real person. Candy.ai’s Community Guidelines prohibit creating or sharing images that resemble real persons, and non-consensual intimate imagery. The same idea applies to any reference image you upload.

Practical steps:

Privacy of generated images

Generated images are part of your account content, so read how the provider handles them. Check:

GirlfriendGPT’s Terms say it can generate images from a text description, but they also say it cannot guarantee that the images will be safe and appropriate. Treat every output as something to review before you keep or share it. Leave personal details such as your address, workplace or identity documents out of your prompts.

Steps to take before you generate

  1. Open the provider’s terms and find the section on tokens, credits or extended features.
  2. Note the cost of one image, one retry and one higher-detail option on the provider’s own pricing page.
  3. Set a spending limit in your own budget before you start.
  4. Read the community guidelines on real people and on intimate imagery.
  5. Test with a fictional description first, then check the storage and deletion settings.

Comparing providers

If you want to see how Candy.ai presents its tokens and image features, you can start on the Candy.ai website. This is a partner link, and this site may earn a commission if you sign up. Check the current terms before you register or pay.

Reading next

For more practical guides on AI companion features, visit the VeryFlirty home page.

Candy.ai is an affiliate partner. A signup or purchase through this link may earn VeryFlirty a commission.

Three questions about it

How can I tell whether an image costs tokens or credits?

Open the provider's pricing page and look for the price of a single image. Check whether it is listed in tokens, credits or as a plan feature, and note what a retry costs.

Can I generate an image that looks like a real person I know?

Candy.ai's Community Guidelines prohibit creating or sharing images that resemble real persons. Check the equivalent rules of any other service, and describe an invented character with its own traits instead.

What should I check about privacy before I keep an image?

Check where images are stored, who can see them, whether you can delete them and what the privacy policy says. Review each output before you keep or share it.