Image generators compared: ChatGPT, Gemini, Midjourney and Firefly

The four image generators most people compare in 2026 are ChatGPT Images, Gemini with Google's Nano Banana models, Midjourney and Adobe Firefly. On October 23, 2026 they differ less in raw quality than in three practical things: whether you can use them for free, how the limits are counted and what the terms let you do with the pictures. This guide compares them from their official help pages and terms, read on October 23, 2026, and explains what owning an image actually means.

The four tools at a glance

Table: image generators compared (official help pages and terms, checked October 23, 2026)

ChatGPT Images

OpenAI says ChatGPT Images is available on all tiers, on the web, iOS and Android, and that images with thinking are available on Plus, Pro and Business, with Enterprise and Edu coming. It can add text, make backgrounds transparent, work in any aspect ratio, start from templates such as a poster or a logo and edit part of a picture: you select an area with the selection tool and describe the change, although OpenAI warns that highlights are not always precise and edits may spill outside the selection. A complex image can take a few minutes. Every image is saved under Images, and deleting the conversation is the way to delete the image. The old DALL·E GPT has been retired.

The limits for free users are not published as a fixed number, so plan for them changing. Our page on ChatGPT pricing compares what each plan includes.

Gemini and Nano Banana

Google's image models in the Gemini app are called Nano Banana. Nano Banana 2 handles requests when the Gemini model is set to Flash or Pro, a faster Nano Banana 2 Lite is used with Flash-Lite, and paid subscribers can redo an image with Nano Banana Pro for more detail, especially text and infographics. Google highlights better text rendering, local edits and character consistency across several images. Downloads are 2K with a paid plan and 1K without one.

Limits are compute-based: they refresh every 5 hours until you reach a weekly limit, and Google warns that images, music and video use a lot of it, video most of all. Its paid plans raise them to 2x (Plus) and 4x (Pro) the standard limits, and Ultra to 5x or 20x the Pro limits. There is also a daily quota of Nano Banana 2 images, and if you hit it you cannot redo more images with Nano Banana Pro. With a personal account you must be 18 or over to edit images, and 13 or over (or the age that applies in your country) to generate them. Our page on Gemini pricing has the plan prices.

Midjourney

Midjourney has no free plan: all four tiers are subscriptions. Basic costs $10 a month with 3.3 hours of fast GPU time, Standard $30 with 15 hours plus unlimited images in Relax mode, Pro $60 with 30 hours and Mega $120 with 60 hours, and an annual plan saves 20%. Extra fast time costs $4 an hour. Two details matter for professional use: Stealth mode, which keeps your images private, exists only on Pro and Mega, so on cheaper plans your images are visible to others. And a company with more than $1,000,000 a year in gross revenue must buy Pro or Mega to use the images under Midjourney's terms.

Adobe Firefly

Firefly works with generative credits: Creative Cloud plans include a monthly allocation, the number of credits depends on the plan and the cost of each action depends on the feature. The same credits cover Generative Fill and Expand in Photoshop, image and video generation in Firefly and text effects in Adobe Express, and you can buy more. Firefly also offers partner models inside the same app, including OpenAI's GPT Image, Google's Gemini, Black Forest Labs' FLUX and Runway, which makes it a single place to try several engines with one bill.

Who owns the image

Owning and copyright are not the same thing. OpenAI's terms, both the European version and the one for other countries, say you own the output and OpenAI assigns you its rights, but also that output may not be unique and other users may receive similar results. Google asks you not to violate others' copyright and to use discretion before publishing what Gemini generates. Midjourney says that anyone who has subscribed at any point is free to use their images in just about any way they want, with the revenue rule above.

Whether anyone can claim copyright over the picture is a separate question. The United States Copyright Office concluded in January 2025 that such outputs are protected only where a human author determined sufficient expressive elements, such as creative arrangement or modification of the output, and that prompts alone are not enough. In practice: for a logo or a product image you plan to protect, edit and compose it yourself, keep your working files and do not assume the raw generation is exclusive to you. Our guide on whether ChatGPT is safe covers what happens to the photos you upload.

Which one to pick

• Quick images and edits with no extra bill: ChatGPT or Gemini, which you may already use for text. Both are in our comparison of the best free chatbot.
• Text inside the image, infographics or diagrams: Gemini's Nano Banana Pro on a paid plan, which Google recommends for text and infographics, or ChatGPT Images, which follows instructions to add text.
• A consistent artistic style and lots of volume: Midjourney from Standard up, with unlimited relaxed generations.
• Work that ends in Photoshop or a design team: Firefly, because the credits and the edits live inside the Adobe apps.

Whatever you choose, the prompt matters more than the tool: describe the subject, the setting, the style, the light and the format, and change one thing at a time. Our prompt engineering basics apply to images as well.