Google Gemini
Chat-based photo editing inside Google's AI assistant: describe the change, the image model makes it.
Upload a photo, type the change, get a new version back: image editing built into ChatGPT.
ChatGPT is an AI chat assistant with an image model built in, so it can create and edit pictures inside a conversation. You upload a photo, describe the change in plain words, and refine the result with follow-up messages. A selection tool lets you highlight one region and describe what should happen there. It can also combine several photos into one and generate images with readable text for posters and mockups.
ChatGPT runs on the web, in desktop apps for Windows and Mac, and in iOS and Android apps. Image editing works on the free plan, with daily caps on how many images you can make. Paid subscription plans raise those caps substantially.
Use ChatGPT if:
Skip ChatGPT if any of these describe you:
Yes. Generating and editing images works on the free plan, with tighter limits on how many images you can make and upload. OpenAI publishes no number for the free limit. Paid plans raise the caps.
There is a selection tool: highlight an area, then describe the change for that region. OpenAI notes that selections are not always precise and edits can extend beyond the highlighted area, so check the whole image afterward.
No guarantee. ChatGPT produces a new AI-generated image rather than changing pixels in place. Faces, text and fine details elsewhere in the picture can come back slightly different.
The web app, desktop apps for Windows and Mac, and mobile apps for iOS and Android. Image features need an internet connection on all of them.
Screen-size images, not the original resolution of your photo. You pick the shape with the aspect ratio control or by asking for it in the prompt. For print files it is a poor fit.
Yes. OpenAI puts C2PA metadata in generated images, which tells platforms that read it the image is AI-made. Since May 2026 it also adds a SynthID watermark, hidden in the pixels rather than in the file's metadata. There is no visible watermark on the image, and OpenAI runs a public page that checks for both.
OpenAI states that content from consumer plans may be used to improve its models unless you turn this off in Data Controls, and that business plans are excluded by default. If the photos are sensitive, check those settings before uploading.
Chat-based photo editing inside Google's AI assistant: describe the change, the image model makes it.