What for: image generation and editing from OpenAI with precise prompt following and legible text, available over the API.

What it is and who it's for

GPT Image is OpenAI's image generation and editing model (gpt-image-1): precise prompt following, good text in frame, available through the API and in ChatGPT. For developers and product teams who want to build solid image generation and editing into a product using the same key as the rest of their OpenAI stack. What sets it apart: strong prompt adherence and native integration with the OpenAI ecosystem.

Key features

Generation and editing (inpainting) driven by a prompt and a mask.
Legible text and precise following of complex instructions.
Available through the Images API and in ChatGPT.
Control over output size, quality and format.
Works with reference images.
One key and SDK shared with the rest of the OpenAI API.

Get started in 5 minutes

Grab an OpenAI key and call generation through the Images API:

curl https://api.openai.com/v1/images/generations \
  -H "Authorization: Bearer $OPENAI_API_KEY" -H "Content-Type: application/json" \
  -d '{"model":"gpt-image-1","prompt":"постер кофейни с надписью OPEN","size":"1024x1024"}'

For edits, use the editing endpoint with a source image and a mask.
Tune size and quality to the task.

When to take it, and when not

✅ Take it if: you're building image generation or editing into a product on the OpenAI stack.
✅ Take it if: prompt accuracy and legible text over an API matter to you.
❌ Skip it → "Midjourney" is better if you want maximum artistic aesthetics with no code.
❌ Skip it → "Adobe Firefly" is better if commercially "clean" training data and IP indemnification are critical.

The honest price

Paid through the API (per image/tokens, depending on size and quality); generation is also available in paid ChatGPT. There is no free API. Exact amounts are on the provider's side and change over time.

Gotchas

No free API — every generation costs money.
Safety filters may reject some requests (faces, brands).
For pixel-precise control over composition it loses to node-based pipelines (ComfyUI).
Check usage rights and the content policy for your specific case.

🤖 Prompt accelerator

"I'm generating and editing images for <PRODUCT/TASK> with GPT Image (gpt-image-1). Help me write a prompt with exact in-frame text, a style and a format, plus an example Images API call in <Python/Node>. For edits, explain how mask-based editing works. Estimate the cost of one generation at the size I need."