Edit and restyle images with FLUX 3, Black Forest Labs' unified multimodal model. Coming soon on Muapi — reserve early access, pay per generation later.
About this model
FLUX 3 Image-to-Image lets you edit, restyle, or extend an existing image using a text instruction and one or more reference images. It is built on the same Self-Flow multimodal architecture that powers FLUX 3 Video and FLUX 3 Action, so edits stay physically and stylistically consistent with the source material — subject identity, lighting, and scene geometry are preserved unless the prompt explicitly asks to change them. This positions FLUX 3 Image-to-Image as the natural successor to FLUX Kontext for high-fidelity, instruction-driven image editing once Black Forest Labs opens general access.
Cost analysis
| Provider | Cost | Notes |
|---|---|---|
| muapi | TBD at launch | Coming soon — early access via Black Forest Labs, general availability pricing not yet published. |
| Fal.ai | Not available | FLUX 3 is not yet available on Fal.ai. |
| Replicate | Not available | FLUX 3 is not yet available on Replicate. |
Coming soon — early access via Black Forest Labs, general availability pricing not yet published.
FLUX 3 is not yet available on Fal.ai.
FLUX 3 is not yet available on Replicate.
* Competitor pricing is estimated based on similar model architectures and usage tiers.
Configuration schema
| Parameter | Type | Description | Default |
|---|---|---|---|
| Prompt | string | Text prompt describing the image. Supports long, richly detailed descriptions. | A hyperrealistic portrait of an astronaut standing on red Martian dunes at golden hour, dust particles catching the light, ultra-detailed suit reflections, cinematic depth of field. |
| Image URLs | array | Reference image(s) to edit or use as visual guidance. | https://d3adwkbyhxyrtq.cloudfront.net/ai-images/186/712345784292/4a8c5c70-abcc-4920-873e-b0e219986453.jpg |
| Aspect Ratio | Enum (8 options) | Aspect ratio of the output image. | 1:1 |
| Resolution | Enum (3 options) | Target resolution of the generated image. | 2k |
Text prompt describing the image. Supports long, richly detailed descriptions.
A hyperrealistic portrait of an astronaut standing on red Martian dunes at golden hour, dust particles catching the light, ultra-detailed suit reflections, cinematic depth of field.Reference image(s) to edit or use as visual guidance.
https://d3adwkbyhxyrtq.cloudfront.net/ai-images/186/712345784292/4a8c5c70-abcc-4920-873e-b0e219986453.jpgAspect ratio of the output image.
1:1Target resolution of the generated image.
2kDeveloper documentation
Upload reference image(s) — supply up to 4 images via images_list.
Write an edit instruction — describe exactly what should change, e.g. "replace the background with a sunset beach, keep the subject unchanged."
Set output parameters — choose aspect_ratio and resolution.
Submit the request once live:
curl -X POST https://api.muapi.ai/api/v1/flux-3-image-to-image \
-H "x-api-key: YOUR_API_KEY" \
-H "Content-Type: application/json" \
-d '{"prompt": "Replace the background with a neon-lit city street at night", "images_list": ["https://example.com/input.jpg"], "aspect_ratio": "1:1"}'
/api/v1/predictions/{id}/result until completed, then download the edited image.Frequently asked
It's Black Forest Labs' instruction-driven image editing mode within the FLUX 3 family, letting you supply reference images plus a text prompt to restyle, extend, or edit an existing image while preserving subject identity.
FLUX 3 Image is rolling out in the coming weeks per Black Forest Labs' announcement. Muapi will enable this endpoint as soon as general API access is available.
The planned endpoint supports up to 4 reference images per request, similar to FLUX Kontext's multi-image editing support.
FLUX Kontext is Black Forest Labs' current-generation image editor. FLUX 3 Image-to-Image is its successor, built on a unified image/video/audio architecture for improved coherence and physical plausibility.
Black Forest Labs has not published FLUX 3 pricing yet. Muapi will update this page with confirmed per-generation pricing as soon as it is announced.
Yes — that is one of its primary intended use cases, using reference images to keep a character's appearance stable while changing scene, pose, or style.