AI image generators
Text-to-image tools turn a written prompt into art, illustrations, product shots, icons or logos, and most will also edit an image you already have. Here are the ones worth your time right now, what each is best at, and how their free tiers and pricing actually work.
The tools
- Midjourney — the strongest default for striking, stylised art and concept work. Runs on the web and in Discord. Paid only (subscription); there is no ongoing free tier.
- ChatGPT (DALL·E / GPT image) — the easiest to steer in plain conversation, and good at readable text inside an image. A limited amount is available on the free plan; more with a paid subscription.
- Adobe Firefly — trained on licensed and public-domain content, so it's the safest pick when commercial usage rights matter, and it's built into Photoshop's generative fill. Monthly free credits, then paid.
- Ideogram — the best at rendering accurate lettering, which makes it a strong choice for logos, posters and typography. Has a free tier.
- Leonardo.Ai — flexible, with fine-grained controls and game-asset workflows. Daily free credits, then paid.
- Stable Diffusion — open source. Run it on your own machine (free, private, unlimited) or through a host. The most control if you're willing to tinker.
Which one should you use?
- Best-looking art with least effort: Midjourney.
- Text inside the image (logos, posters): Ideogram.
- Commercial safety / editing real photos: Adobe Firefly.
- Free and private on your own hardware: Stable Diffusion.
Before you publish or sell: check each tool's licence and
commercial-use terms — they differ a lot, and a free tier often grants fewer rights than a paid one.
Regenerate or edit anything that too closely resembles an existing artist's work or a trademark.
See all AI tools by type, or what you can make on the examples page.