GPT Image 2

GPT Image 2 is OpenAI’s state-of-the-art image generation model. Create images from text or edit existing images with precise, instruction-following control.

What it does

GPT Image 2 handles two workflows: generating images from text descriptions, and editing existing images with specific instructions. It’s designed to follow your directions closely while keeping the parts you want unchanged.

When you pass reference images, GPT Image 2 processes them at high fidelity automatically. There’s no knob to adjust — the model always does its best to preserve the details of the input. Pass one image to edit it, or pass multiple images to combine styles, subjects, or references into a single output.

Key capabilities

Photorealism and detail: Natural-looking images with accurate lighting, believable materials, and rich textures. From honest, unposed photography to polished commercial visuals.

Text rendering: Dense text, small lettering, and complex layouts like infographics, UI mockups, and marketing materials.

Precise editing: Targeted changes without reinterpreting the entire image. The model preserves identity, composition, and lighting while you adjust specific elements.

Style control: Apply consistent visual styles across different subjects, or transfer the look of one image to another with minimal prompting.

World knowledge: Ask for a scene set in “Bethel, New York in August 1969” and the model understands you want Woodstock — it has reasoning built in.

Use cases

Image generation: Infographics, logos, UI mockups, photorealistic scenes, comic strips, marketing visuals. Works well for both creative exploration and production-ready outputs.

Image editing: Style transfer, virtual clothing try-ons, product mockups, text translation in images, lighting adjustments, object removal, scene compositing. Insert people into new scenes while preserving their likeness, or swap furniture in room photos without changing the camera angle.

Character consistency: Build multi-page illustrations where characters look the same across different scenes — useful for children’s books, storyboards, and campaigns.

How to get good results

Be specific: Describe what you want clearly. Instead of “make it better,” say “add soft coastal daylight” or “change the red hat to light blue velvet.”

Use photo language for realism: Mention lens type, lighting quality, and framing when you want photorealistic results. “Shot with a 50mm lens, soft daylight, shallow depth of field” gets you closer to real photography than generic quality terms.

Lock what shouldn’t change: When editing, explicitly state what must stay the same. “Change only the lighting, preserve the subject’s face, pose, and clothing” prevents unwanted alterations.

Put text in quotes: For readable text in images, put the exact copy in “quotes” and describe the typography. “Bold sans-serif, centered, high contrast” helps ensure legibility.

Iterate with small changes: Start with a base image, then make one adjustment at a time rather than rewriting everything.

Reference multiple images clearly: When working with several input images, label them by number and describe how they relate. “Apply the style from image 1 to the subject in image 2.”

Inputs

• prompt: What you want to generate or how to edit the input
• input_images: One or more reference images (for editing or composing)
• aspect_ratio: 1:1 (square), 3:2 (landscape), or 2:3 (portrait)
• quality: low, medium, high, or auto — lower quality is faster and cheaper
• number_of_images: Generate up to 10 images in a single call
• output_format: webp (default), png, or jpeg
• background: auto or opaque
• moderation: auto (default) or low for less strict content filtering
• openai_api_key: Optional — bring your own OpenAI API key to pay OpenAI directly

Notes

GPT Image 2 doesn’t support transparent backgrounds. For transparent PNGs, use openai/gpt-image-1.5.

You can try this model on the Replicate Playground at replicate.com/playground.