About GPT Image 3.0
GPT Image 3.0 is a browser workspace for OpenAI image generation. You write a prompt, optionally add your own reference photos, and download a finished image. The site runs the current OpenAI image model today and will switch to GPT Image 3 in the same workspace when OpenAI releases it.
Last updated
How the workspace is built
- Prompts and reference photos go to the current OpenAI image model through our own server; the browser never holds an upstream key.
- Every generation is charged in credits at the resolution you choose, and failed generations return their credits.
- Model names are shown as they are. The generator labels the model that actually produced each image, including the examples on the home page.
What you can do here
- Generate from text, with control over aspect ratio and output resolution up to 4K.
- Edit with one or more reference photos and keep a face, product or layout intact.
- Render headlines, labels and short paragraphs inside the image, and use the same workspace from a browser or the API.