Short answer
Don’t choose GPT Image 1 for a new OpenAI API integration. OpenAI says it was shut down on October 23, 2026, and is no longer available. If you already use it, plan a migration rather than treating the old model ID as a supported production dependency. (developers.openai.com)
GPT Image 1 was OpenAI’s image-generation model, identified in the API as gpt-image-1. It supported generating images from prompts and editing images using reference images and masks. Those capabilities may explain why it suited some workflows, but they do not make it a viable API selection now that the model has been retired. (developers.openai.com)
What GPT Image 1 did
GPT Image 1 handled image generation and editing. OpenAI’s retired-model prompting documentation describes using reference images and masks for these tasks. That made it relevant to workflows such as creating an image from a prompt or asking for a change to an existing image. (developers.openai.com)
For production selection, the key distinction is its lifecycle, not a comparison of its former strengths with today’s models: GPT Image 1 is shut down. Don’t start a new integration against gpt-image-1, and don’t assume an existing request will continue working just because older code or third-party guides still document it. (developers.openai.com)
What to evaluate instead
OpenAI’s current image prompting guidance identifies two GPT Image 2.6 options: GPT-Image-2.6 Sunburst (gpt-image-2.6-sunburst) for workflows that need more detail or precise edits, and GPT-Image-2.6 Flare (gpt-image-2.6-flare) when speed matters most. OpenAI also documents GPT Image 2.5 models in its image-generation materials, so check the current API reference and your account’s model availability before settling on a model ID. (developers.openai.com)
| If your workflow prioritizes… | What to evaluate |
|---|---|
| Detailed compositions or precise edits | Start by evaluating GPT-Image-2.6 Sunburst. |
| Faster image iteration | Start by evaluating GPT-Image-2.6 Flare. |
| A single image-generation or edit request | The Image API. OpenAI describes it for generating from a prompt or editing an existing image. |
| Conversation-based or iterative image editing | The Responses API, which OpenAI describes for image generation within multi-step conversations and iterative edits. |
These are documented starting points, not a guarantee that a model will meet your quality, latency, or cost targets. Test the models available to your account on representative prompts and reference images before choosing. OpenAI’s docs describe the Image API and Responses API as different options for these workflows. (developers.openai.com)
Migration checklist for existing API users
- Find every use of
gpt-image-1. Check application code, deployment settings, queued jobs, and saved request templates. - Choose a currently available model and API path. OpenAI documents image generation through the Image API and as a tool in the Responses API; choose based on whether you need a standalone generation/edit request or image work inside a multi-step conversation. (
developers.openai.com) - Recheck request fields and image handling. Don’t carry over old parameters, output assumptions, or image-input limits without confirming that they apply to the replacement model and endpoint. Supported options depend on the model and API path. (
developers.openai.com) - Run a task-specific comparison. Reuse a small set of real prompts, reference images, and edit instructions. Review the outputs for details your product depends on—such as text, layout, or preservation of unchanged parts of an image—and measure response time and cost using the current model’s documentation and your own requests.
- Update monitoring and failure handling. Confirm your application handles the replacement model’s actual responses and errors; then monitor successful generations, latency, and spend during rollout.
OpenAI says developers may need to complete API Organization Verification before using GPT Image models. Treat that as a potential access prerequisite and confirm it in your developer account. (developers.openai.com)
Pricing: don’t reuse GPT Image 1 estimates
Older third-party articles quote GPT Image 1 rates and per-image estimates, but those figures are not a useful budget for a model that is no longer available. Price the replacement using the current OpenAI pricing and image-generation documentation, with the model, output settings, and expected input images you plan to use. Don’t treat a historical per-image estimate as a current quote. (developers.openai.com)
Bottom line
GPT Image 1 is a retired model, not a sensible choice for a new production API integration. If you have an existing integration, migrate it. For a replacement, OpenAI’s current prompting guidance points to Sunburst for more detail or precise edits and Flare when speed matters most. Compare available models on your own tasks, verify access and request details in the current API reference, and estimate cost from the replacement model’s current pricing—not from GPT Image 1’s historical rates. (developers.openai.com)