GPT-Talk

GPT Image 2 complete guide: text-to-image, editing, style control

How to get reliable, high-quality output from GPT Image 2 and the Images API: prompt structure, style control, image editing, and cost.

TL;DR
GPT Image 2 is OpenAI's flagship text-to-image model — a major quality jump over its predecessors. Prompting, style control, and editing each have their own gotchas. This handbook covers the prompt framework, style keywords, inpainting, and pricing.
GPT Image is OpenAI's text-to-image model family. GPT Image 2 is available through both ChatGPT and the Images API and supports generation, edits, and variations.

How to

  1. Write a structured prompt

    Use the "subject + style + lens/aperture + background + aspect" five-part form: "An orange tabby cat on a windowsill, photorealistic, shallow depth of field, soft-lit background, 3:2."

  2. Call the Images API for your first generation

    POST /v1/images/generations with model="gpt-image-2", prompt=..., size="1024x1024", quality="medium", n=1. The response carries b64_json.

  3. Pick a style: natural (realistic) vs vivid (dramatic)

    Vivid leans dramatic with saturated colours; natural is restrained and realistic. Run the same prompt with both values and pick.

  4. Use the edits endpoint for inpainting

    POST /v1/images/edits with the original image plus a same-sized PNG mask (transparent area = repaint) and a prompt; the model only repaints the masked region.

  5. Pre-launch Usage Policy self-check

    Pull the latest prices from openai.com/api/pricing. Don't generate real people, unauthorised brands, or disallowed content. Preserve provenance metadata on commercial assets.

GPT Image 2 complete guide: text-to-image, editing, style control

GPT Image is OpenAI's text-to-image model family. GPT Image 2 is available through both ChatGPT and the Images API and supports generation, edits, and variations.

1. Overview

GPT Image 2 is OpenAI's flagship text-to-image model — a major quality jump over its predecessors. Prompting, style control, and editing each have their own gotchas. This handbook covers the prompt framework, style keywords, inpainting, and pricing.

2. Key points

  • GPT Image 2 prompt formula: subject + style + lens/aperture + background + aspect — be specific
  • Two styles: natural (photorealistic) and vivid (more dramatic and artistic)
  • Image edits take a mask for inpainting; variations generate alternate takes of an input image
  • Images API supports size (1024x1024, 1024x1536, 1536x1024, or 'auto'), quality (low/medium/high), and n (count)
  • Pricing is per output token (~$40 per 1M); higher quality and larger sizes cost more — check the Usage Policy before commercial use

3. How it works

The sections below unpack each point. Skim alongside the OpenAI documentation for full context.

4. Practical steps

  1. Define 'done' up front — what does success look like?
  2. Pick the right model using the comparison above.
  3. Run the minimal example end to end; record parameters and the model version.
  4. Integrate into your existing code.
  5. Monitor logs and failure modes; review weekly.

5. Common errors and fixes

  • 401: invalid or expired API key — rotate it.
  • 429: rate limit hit — enable exponential backoff retries.
  • 400: bad request — verify the model name and messages shape.
  • Poor quality output: try a different model, add few-shot examples, trim the prompt.

6. Next steps

Once you are comfortable here, continue with:

  • ChatGPT Complete Guide (2026): From Beginner to Expert
  • Prompt Engineering Core Patterns: 8 Templates That 2x GPT Output
  • OpenAI API Beginner: Your First GPT-5.6 Call

Key points

  • GPT Image 2 prompt formula: subject + style + lens/aperture + background + aspect — be specific
  • Two styles: natural (photorealistic) and vivid (more dramatic and artistic)
  • Image edits take a mask for inpainting; variations generate alternate takes of an input image
  • Images API supports size (1024x1024, 1024x1536, 1536x1024, or 'auto'), quality (low/medium/high), and n (count)
  • Pricing is per output token (~$40 per 1M); higher quality and larger sizes cost more — check the Usage Policy before commercial use

Frequently asked questions

GPT Image 2 replaced the DALL·E 2/3 family on 2026-05-12 and is now OpenAI's only text-to-image model. It produces noticeably better text rendering, more accurate prompt adherence, and stronger multi-subject composition than DALL·E 3, and it is the only model exposed by both ChatGPT and the Images API going forward.

Official references

Subscribe to GPTMap Weekly

One email every Monday: curated OpenAI updates, deep dives, and best practices. No ads, unsubscribe anytime.

GPTMap EditorialPublished 2026-07-12Updated 2026-07-14 2 min read
Test environment (EEAT)
Last tested: 2026-07-14
Model used: gpt-5.6