GPT Image
OpenAI's image model follows long prompts closely, renders text well and edits photos from instructions.
- Detailed scenes
- Text in images
- Editing
Made with GPT Image
Real output from our prompts, unedited.
“Candid street photograph of an elderly fisherman mending a bright orange net on a harbour wall at golden hour, 85mm lens, shallow depth of field, natural skin texture.”
“Retro travel poster for a fictional island called "Kwirl Bay". Headline "KWIRL BAY" in bold art-deco letters, smaller line "Sun all year", turquoise sea, coral cliffs, flat screen-print style.”
“Turn this into a calm blue-hour scene: warm lanterns glowing on the boat, first stars in the sky, keep the boat and lettering exactly as they are.”
“Overhead photo of a breakfast table: sourdough toast with avocado, two flat whites, a folded newspaper, morning sun through linen curtains.”
“Hand-lettered chalkboard menu for a cafe called "The Little Kwirl" listing Flat white $5, Chai latte $5.50, Banana bread $6, with small doodles.”
“Make it a rainy afternoon: wet windows, grey light, a small umbrella leaning on the table. Keep the food the same.”
Specs
OpenAI image model with excellent prompt adherence and editing.
- Aspect ratios
- 1:1, 3:2, 2:3
- Images per request
- Up to 4
- Edits a reference image
- Yes
- Negative prompts
- No
Compare with
GoogleNano Banana Pro
Google's most detailed image model, photoreal and great at text.
GoogleNano Banana
Gemini image model for conversational edits and consistency.
Black Forest LabsFLUX1.1 Pro Ultra
Black Forest Labs' high-resolution photographic model.
Black Forest LabsFLUX Kontext Pro
Generate or edit images with text instructions.