Meta Muse Image Edit is the editing counterpart to Meta Muse Image, from Meta Superintelligence Labs. Point it at an existing photo and describe the change: the model understands what's already in the frame and scopes edits to just what you asked for, or blends in multiple reference images while keeping the rest of the picture intact. Like its generation sibling, it reasons through the instruction before acting and reviews its own draft, rather than mapping a prompt straight to pixels.
| Developer | Meta (Meta Superintelligence Labs) |
|---|---|
| Model ID | meta/muse-image-edit |
| Release Date | August 26, 2026 |
| Modalities | Text, Image (input) → Image (output) |
| Pricing | $0.013 per image |
Understands the contents of the input image and reasons over a multi-part instruction before it renders the change.
Changes are scoped to what you asked for — the rest of the image is left intact instead of being regenerated wholesale.
Blend in multiple reference images — people, objects, styles, environments — into a single edited result.
Reviews its own draft inside its chain of thought, making a local correction or a fuller regeneration when needed.
Reads and renders text accurately within edits — useful for labels, signage, and UI mockups.
Ranks in the top tier of human-preference leaderboards for single- and multi-image editing at launch.
Swap backgrounds, adjust details, or restyle product photos while keeping the rest of the shot untouched.
Combine a subject from one photo with a style or setting from another, blended into a single coherent edit.
Make a series of scoped, local edits on the same base image without regenerating the whole picture each time.
Measured on the AI/ML API gateway across live traffic. Figures are medians unless stated. Last updated 2026-10-12.
Edit an existing image with a text instruction, changing only what is asked while keeping the rest intact. It is suited to removing objects, swapping backgrounds, and restoring old pictures from a photo upload.
It is also suited to combining multiple reference images into one coherent scene. The model is documented for multi-reference composition, such as placing people, products, or environments from separate photos into a single image.
import requests r = requests.post( "https://api.aimlapi.com/v1/images/edits", headers={"Authorization": "Bearer " + AIMLAPI_KEY}, data={"model": "meta/muse-image-edit", "prompt": "Add a crown to the cat's head."}, files={"image": open("input.png", "rb")}, ) print(r.json())
const form = new FormData();
form.append("model", "meta/muse-image-edit");
form.append("prompt", "Add a crown to the cat's head.");
form.append("image", fileInput.files[0]);
const r = await fetch("https://api.aimlapi.com/v1/images/edits", {
method: "POST",
headers: {
Authorization: `Bearer ${process.env.AIMLAPI_KEY}`,
},
body: form,
});
console.log(await r.json());
curl -L \ --request POST \ --url 'https://api.aimlapi.com/v1/images/edits' \ --header 'Authorization: Bearer <YOUR_AIMLAPI_KEY>' \ --form "model=meta/muse-image-edit" \ --form "prompt=Add a crown to the cat's head." \ --form "image=@<YOUR_IMAGE_PATH.png>;type=image/png;filename=<IMAGE_NAME>"
OpenAI-compatible — swap the base URL and it works with your existing SDK.
| Type | Price |
|---|---|
| Output | |
| Model | Input | Output | Context | Best for |
|---|---|---|---|---|
Meta Muse Image Edit This page | Reference-guided image editing | |||
| Image generation at scale | ||||
| Image generation at scale | ||||
| Image generation at scale | ||||
| Image generation at scale |
Meta Muse Image Edit takes text, image as input and returns image.
Meta Muse Image Edit became available on August 26, 2026.
Meta Muse Image Edit is priced at $0.013 / gen.
Meta Muse Image Edit is billed per generation — a fixed charge per output rather than by prompt length.
Yes, Meta Muse Image Edit accepts image input as well as a text prompt.
Meta Muse Image Edit was built by Meta.
Send a request to https://api.aimlapi.com/v1/images/edits with meta/muse-image-edit as the model id.
Yes. Meta Muse Image Edit is served through AI/ML API, so the same key and endpoint format used for other models applies.
Yes, it can compose an edit using multiple reference images.
Yes, it stays coherent across multiple turns of editing.
