GPT Image 1 Mini Edit Image to Image: Photo Variants

Last verified: October 19, 2026

OpenAI recommends GPT Image 1 Mini for quick, high-volume image-to-image work: rapid ideation, previews and draft assets that don't need its strongest editing performance. The model takes text and images as input and returns an image, so your photo and a written instruction are the whole brief.

GPT Image 1 Mini runs on OpenAI's image edits endpoint, which takes source images and a prompt and returns an edited or extended version. Use the prompt to decide how far the result moves away from your upload. Ask for one narrow change to stay close to the original, or describe a whole new scene to use your files as references.

Vidofy keeps the form to three decisions: which photos go into Upload Images, what the Prompt says, and which Output Format you download. With so few settings, the prompt does most of the work. Name the change, list what must stay and give each upload a role. The tips below show how to word each of these for this model.

Explore more image to image

Image To Image Seededit V3 Nano Banana 2 GPT Image 2 Seedream 4.5 Edit Qwen Image 3.0 Qwen Image 3.0 Pro
Capability Snapshot

GPT Image 1 Mini editing at a glance

What you can upload, set and get back from each run.

Input

Up to 8 images (JPG, JPEG, PNG or WEBP, up to 10 MB each) + a text prompt of up to 2,048 characters

Output Format

JPEG or PNG

Output

1 image per run

Prompting tips for better GPT Image 1 Mini edits

Separate what changes from what stays

Write the edit in two parts in the Prompt box. First comes a 'change only' line naming the one thing to alter. Then add a keep-list of what must not move: identity, geometry, layout, camera angle, brand elements. Close with explicit exclusions such as no watermark, no extra text or no logos when the source is clean. This change-plus-preserve pattern is what OpenAI's prompting guide for its GPT Image models recommends for edits. It also gives you a checklist to judge the result against.

Give a style reference a single job

When an upload is there only for its look, tell the model which part of it to borrow (its palette, texture or visual medium) and describe the new subject separately. With several uploads, label each one by its order and role, for example 'Image 1: product photo; Image 2: style reference'. Then say how they interact, for example 'apply Image 2's palette to Image 1'. OpenAI notes that outputs can differ across its models, so test the wording on a single run before building a batch on it.

Describe the finished image in a fixed order

OpenAI's prompting advice is to start from the image you need, then cover subject, composition, style and constraints. In practice, open the Prompt with what the final picture is and what it's for. Then name the viewpoint, lighting and materials before listing restrictions. A brief that describes the end result is easier to check the output against than a bare command like 'make it better'. Keeping the same order every time also makes a long prompt quick to trim when you run short of space.

How to edit images with GPT Image 1 Mini

Five steps from your first upload to a downloaded edit.

GPT Image 1 Mini or GPT Image 2 for image edits

How the two models split editing jobs, based on OpenAI's and Microsoft's model documentation.

Kind of work each is pointed at OpenAI positions GPT Image 1 Mini for high-volume, lower-stakes output such as drafts, previews and variant batches that don't need the strongest editing performance. OpenAI's guide steers customer-facing assets, photorealistic generation and editing-heavy workflows toward GPT Image 2. Use GPT Image 1 Mini while you are still exploring directions, and move to GPT Image 2 when a single edit is the deliverable a client will see.
Keeping a person's face recognizable Microsoft's Azure model table lists no dedicated face preservation for GPT Image 1 Mini and calls it better suited to non-portrait, general creative imagery. The same table credits GPT Image 2 with advanced face preservation for realistic, consistent results. Choose GPT Image 2 for portrait retouching or likeness-critical edits, and keep GPT Image 1 Mini for products, scenes and illustration.
Turnaround across many runs Microsoft's table describes GPT Image 1 Mini as faster for large-scale or iterative generation. GPT Image 2 is described there as a high-fidelity, realism-optimized model with higher latency. Pick GPT Image 1 Mini when you plan many quick passes, and GPT Image 2 when you can wait longer per result for fidelity.
Start drafts, variant sets and non-portrait edits with GPT Image 1 Mini. Take portrait-critical or client-facing final edits to GPT Image 2.

Common GPT Image 1 Mini editing problems and fixes

Problems that come up when editing with GPT Image 1 Mini, and what to change before the next run.

The person in my edited photo doesn't look like the same person anymore

Microsoft's Azure documentation lists GPT Image 1 Mini without dedicated face preservation and as better suited to non-portrait, general creative imagery, so be careful with portrait edits. In Prompt, state which facial features and proportions must remain unchanged. Keep the requested change away from the face where you can, then run it again. If likeness is the whole point of the job, the model comparison above covers a better fit.

Details I wanted to keep shifted after a few rounds of edits

Repeated edits can still change details you meant to preserve. Paste the same keep-list into Prompt on every run, ask for one change at a time and compare each download with the one before it. When something has slipped, upload the last good file again rather than correcting on top of the drift, then run again.

Words in the image come out misspelled or in the wrong spot

Precise text placement and clarity are a known weak spot for OpenAI's GPT Image models. Keep any added wording short, write the exact words in quotes and say where they sit (for example 'small cream tag, lower right'), then run again. If the typography must be letter-perfect, add it in a design tool after downloading.

My edit is taking a long time to come back

Complex prompts can take up to two minutes to process, so give the run time before trying again. For draft rounds, cut the Prompt down to the change and its keep-list, and choose the faster file type described in the Output Format step.

My photo won't upload

Upload Images only takes the file types, file sizes and minimum dimensions listed in the upload FAQ below, and it caps how many files a run can hold. Convert other formats to one of the accepted types. Export a lighter copy if the file is too large, use a bigger original if it is too small, and remove extra files beyond the cap. Then add the image again.

Frequently Asked Questions

What image files can I upload to GPT Image 1 Mini?

Upload Images accepts JPG, JPEG, PNG and WEBP files of up to 10 MB each, at least 300×300 pixels in size. Each run needs at least one image and takes up to eight, so beyond the photo you're editing, the rest can serve as style or content references.

How many images does each run return?

Each run returns one image. To compare alternatives, run again with the same uploads and change one thing in Prompt each time, so the differences between versions stay easy to read.

How is the cost of a run shown?

The credit cost of a run is calculated from the options you choose. Once you're signed in, it appears above the Generate button.

Can I edit the result again?

Yes. Both file types offered under Output Format are formats that Upload Images accepts, so a downloaded result can go straight back in as the source for the next run. Write the new prompt around the next single change and carry over the details you want kept.

References

Sources and citations used to support the content provided above.

Updated: 2026-10-21 15:46:05 6 Sources

developers.openai.com

Source Link
https://developers.openai.com/cookbook/examples/multimodal/image-gen-models-prompting-guide

developers.openai.com

Source Link
https://developers.openai.com/api/docs/models/gpt-image-1-mini

developers.openai.com

Source Link
https://developers.openai.com/api/reference/python/resources/images/methods/edit

developers.openai.com

Source Link
https://developers.openai.com/api/docs/guides/image-generation

developers.openai.com

Source Link
https://developers.openai.com/api/docs/guides/image-prompting

learn.microsoft.com

Source Link
https://learn.microsoft.com/en-us/azure/foundry/openai/how-to/dall-e