You do not need a subscription just to make an illustration or an invitation. Compare five tools for creating AI images for free and the limits of their free plans.
You need an illustration for an article, a background for an invitation or an image for a presentation. AI can create one from a short description, and a free account is often enough for your first attempts. Finding the generate button, however, can be easier than working out what you actually get for free.
Some tools make you wait after a few attempts. Others make your images public or charge for further edits. Your choice also depends on what you plan to do with the image afterwards.
How we chose the tools
We selected five tools that work in a browser and let you generate images without a paid subscription. For each one, we considered:
- What is actually free. Is there an ongoing free plan, or just a short trial of a paid subscription?
- What you can use it for. Creating an illustration, editing a photo and designing an image with lettering are different tasks.
- Where the restrictions are. Generation limits, editing features, downloads and whether outputs are public.
- What the provider says about using the images. A free account and permission for commercial use are separate matters.
We checked official pricing pages and guides on 5 September 2026. For Grok's segments, we also link to another author's hands-on example. This selection is based on those sources; it is not our own image-quality test using identical prompts. We therefore do not score the tools or declare one the best. Where limits change, we link to the current rules rather than promise a fixed number of images per day.
Which image generator to try
If you already use ChatGPT, Gemini or Grok, start there. You do not need another account for your first illustration. Grok is also worth considering if you want to select an individual element from an image and work with it separately. Try Ideogram for a design with lettering, or Firefly if you want to work with Adobe tools.
| Tool | When to try it | What is free | Main restriction |
|---|---|---|---|
| ChatGPT | You want to refine an image through a conversation | Image generation with usage limits | Image generation has its own limit; once you reach it, you have to wait |
| Gemini | You want to generate an image or edit your own photo | Generation and editing without a Google AI subscription | Limits change; availability also depends on your account |
| Grok Imagine | You want to select individual objects for editing or separate downloads | Image generation on Free; check access to specific editing features in your account | Outputs carry a watermark; video generation requires a paid plan |
| Ideogram | You are trying a poster, cover or lettering within an image | Weekly credits for eligible accounts; the amount may vary | Images created on the free plan are public |
| Adobe Firefly | You want to create images within Adobe's tools | Limited daily generations without a subscription | Access depends on the feature and selected model |
Credits are not the same as a number of images. One prompt can produce several versions, and a more demanding model can use more credits. The sections below explain the details and link to the sources.
ChatGPT: refine the result through a conversation
In ChatGPT, you can start with a simple description and add instructions afterwards. Ask for a lighter background, an object to be removed or more space for a headline. This is useful when you are still working out what you want. The ChatGPT image guide explains the editing options.
The free account includes image generation, but it has a separate usage limit. You may still be able to continue a text conversation even when you cannot create another image. Once you reach the limit, you have to wait for it to reset. See the ChatGPT Free Tier FAQ for details.
Check the parts you wanted to keep, too. Saying “change only the background” does not guarantee that the shape of a cup or a detail on someone's clothing will stay the same.
Gemini: create images and edit your own photos
Gemini lets you create an image from a prompt, edit an uploaded photo or combine several images into a new composition. If you already have a Google account, it is another option you can try without a new subscription. Google's guide to creating and editing images in Gemini explains the process.
Image generation is available without a paid Google AI plan. Google does warn that limits can change with service capacity. Access may differ for work or school accounts. Do not rely on older guides that promise everyone the same number of images.
For Nano Banana 2, Google lists downloads at 1K resolution without a subscription and 2K with a paid plan. Image editing is available to users aged 18 or over. Changing the background of your own photo of an object is a useful first task to try.
Grok Imagine: work with individual parts of an image
Grok Imagine is interesting for what you can do with an image after generating it. Image 2.0 recognises individual elements and offers them in a sidebar as segments. Select a cup on a table or an icon in an illustration, then describe the change you want. You can also download a selected element separately, as shown in this author's hands-on example of the segments panel.
This is useful when you like just one object in a generated design and want to use it elsewhere. Think of segments as recognised areas you can select and edit. They do not, by themselves, mean you get a file with layers and editable text like those in a graphics editor.
The Imagine Image 2.0 announcement also confirms precise area selection, background removal with transparent output and aspect-ratio changes that fill in the scene. You can remove an unwanted object with a written instruction, such as “remove the cable from the desk”. Check outside the selected area as well, because other details can change during an edit.
The official Grok plan comparison includes image generation on Free, without specifying a fixed number of generations. We have not confirmed that all the editing features described here are available on a free account. The comparison lists video generation under paid plans only.
Expect a Grok watermark on the output. According to the image and video FAQ, it cannot be disabled in the app and the service's rules prohibit removing it.
Ideogram: when lettering is part of the design
Ideogram also focuses on text and typography within images. Try it for a café name on a shop window or a short line on a poster. Its own guide still warns about possible spelling mistakes and missing letters or words.
According to Ideogram's current pricing page, eligible free accounts receive weekly credits when signing in through Google, Apple or Microsoft. The amount may vary. Expect slower generation; AI editing and image references require paid priority credits.
Privacy may be the more important restriction: images generated on the free plan are public. That makes it unsuitable for an unreleased client campaign. Public output does not mean commercial use is prohibited, though — Ideogram allows it under its terms. Both rules appear in the FAQ below the pricing table.
Adobe Firefly: when you work with Adobe tools
Consider Firefly if you want to create backgrounds or illustrations and continue working on them in Adobe's tools. Its current pricing page lists limited daily generations without a paid plan. These are separate from the trials of paid subscriptions offered on the same page.
When choosing a model, check who provides it. Firefly gives access to Adobe's own models and models from other companies. Adobe's statements about the training data for its own models cannot automatically be applied to everything available in the app.
In its Firefly FAQ, Adobe permits commercial use of outputs from features without a beta label. Beta outputs can also be used commercially unless the specific feature states otherwise. That does not guarantee that you can use every generated image regardless of its content.
How to write an image prompt in plain English
You do not need a long list of specialist terms to get started. Describe what should be in the image, where the scene takes place and how you intend to use the result. If the tool has an aspect-ratio selector, set it in the interface too.
“Create a nice image about coffee” leaves almost everything to the generator. A more useful description might look like this:
Create a landscape image for an article about making coffee at home. On a wooden kitchen table are a hand grinder, a ceramic cup and a few coffee beans. Morning light comes through a window on the left. Make it look like an everyday photograph with natural colours. Leave the right third clear for a headline. No text or logos, aspect ratio 16:9.
If the result is wrong, name the specific problem. “Make it nicer” does not explain much. Try this instead:
There are too many objects on the table. Keep only the cup and grinder, and remove everything else. Keep the camera angle and lighting.
Change one thing at a time. You will see which adjustment helped and avoid spending all your free attempts on completely new designs. For more examples of clearer instructions, see 25 practical ChatGPT prompts in English.
Add the final text in an editor
It is worth trying generation for a short piece of lettering. For an invitation with a date, address and names, a more reliable approach is to generate the background without text and add the details yourself. One wrong digit can spoil the whole invitation.
You can add a headline with Canva's text editor, for example. Choose the font, fix any spelling and change the date without generating the image again. Canva can finish the design, so you do not also need to try its AI generator for this task.
What to check before using the image
Open the image at full size, not just as a thumbnail. Check hands, faces, object edges and any lettering. For a product, compare the result with a real photo. Check the downloaded file's dimensions and any watermark too.
Before publishing on a business website or in an advertisement, check the terms for the particular plan and model. Permission from the provider does not resolve rights to another person's photo, logo or likeness that you supplied in your prompt. A paid account does not replace that check.
For your first attempt, choose one real task, such as an illustration for an article you are preparing. Write a prompt, make a few focused edits and place the result where you intend to use it. Only then will you know whether the tool saved you work and is worth returning to.
