# GPT Image API

> GPT Image API for text-heavy GPT Image 2 generation and editing — posters, menus, app mockups, packaging comps, and social cards with reviewable copy.

GPT Image API is the image stack teams reach for when the copy inside the image matters as much as the pixels around it. Prompt patterns, image edit workflows, multilingual review guidance, flexible output controls, and async task handling are documented below and in the developer docs.

## Public Pages

- [GPT Image API](https://gptimageapi.dev/): GPT Image 2 generation and editing, flexible output controls, and prompt patterns for teams shipping text-heavy posters, menus, mockups, packaging comps, and social cards.
- [GPT Image API System Announcements](https://gptimageapi.dev/announcements): Official notices about model availability, service changes, and promotions.

## Reference

- [API documentation](https://gptimageapi.dev/docs): One-page developer guide — quick start, authentication, endpoints, request fields, responses, errors, and code examples.
- [Full AI-readable product reference](https://gptimageapi.dev/llms-full.txt): Detailed capabilities, access rules, workflows, and canonical links.
- [XML sitemap](https://gptimageapi.dev/sitemap.xml): Canonical public URLs and localized alternates.
- [Contact](mailto:support@gptimageapi.dev): Product support and factual corrections.

## Prompting Guidance

- Wrap literal copy in quotes and keep each text region short so the model treats it as non-negotiable content.
- Specify the layout context clearly, and separate must-have text from descriptive styling instructions.
- Use seeded reruns for consistent team review; store prompts and seeds alongside each request.
- For Chinese, Japanese, and Korean (CJK) text, keep literal copy concise and validate spelling, placement, and legibility with representative inputs before publishing.
- Prefer image edit routes (`POST /v1/images/edits`) for copy, packaging, or scene updates on brand-approved assets instead of regenerating from scratch.

## Common Workflow Hints

- To generate an image, call `POST /v1/images/generations` with `model: "gpt-image-2"` and read the image URL from `response.data[0].url`.
- For long-running image tasks, use `POST /v1/async/images/generations` and poll `GET /v1/tasks/{id}` until `status` is `completed`, then read the asset URL from `result.data[0].url`.

## Optional

- [Privacy Policy](https://gptimageapi.dev/privacy-policy): How GPT Image API handles personal information and service data.
- [Terms of Service](https://gptimageapi.dev/terms-of-service): Terms governing use of GPT Image API.