- Text-to-image and edit
Modes
- Up to 16 for edits
Input images
- 20 MP, e.g. 5160 × 3870
Max size
- Any from 1:3 to 3:1
Aspect ratios
About MiniMax H3 Image
MiniMax H3 Image is the still-image member of MiniMax's H3 family, whose video side Masonry offers as H3 Max. fal launched it on November 10, 2026, and Masonry runs it through two fal endpoints. Text-to-image creates a new image from a prompt, and fal describes it as following instructions precisely across photorealistic scenes, illustrations, and visual designs. Edit takes one to sixteen reference images and an instruction, then changes one image or combines several into a new one, with each input addressed in the prompt as Image 1, Image 2, and so on.
Two limits stand out. Sixteen references in a single request is the most of any image model in Masonry, which suits compositions that need a person, a product, a setting, and several supporting details at once. And the output size is open: any width and height from about 0.66 to 20 megapixels, with each edge between 468 and 7,745 pixels and a shape between 1:3 and 3:1. With no size set, a new image is 2048 × 2048, and an edit follows the shape of its first reference. This page describes the model as fal documents it; Masonry has not published its own test results for it yet.
Why teams choose MiniMax H3 Image
Choose MiniMax H3 Image when one image has to draw on many references, or when the output needs a size or shape other models do not offer. Sixteen references is more than FLUX 3 Image's ten or Krea 3's eight, and 20 megapixels is well above a 4K frame. It has no seed, no negative prompt, and no quality setting, so control comes from the prompt and the references. Masonry has not benchmarked it against the other image models yet, so compare it on your own source images before moving a workflow to it.
What MiniMax H3 Image can do
The capabilities that set MiniMax H3 Image apart and earn its place in a brief
Combine Up to 16 Images
The edit endpoint accepts one to sixteen reference images in one request. Refer to them as Image 1, Image 2, and so on, in the order you attach them, to say which is the product, which is the person, and which only sets the scene.
Custom Sizes Up to 20 Megapixels
Request any width and height between 655,360 and 20,000,000 pixels, with each edge from 468 to 7,745 pixels and an aspect ratio from 1:3 to 3:1. Presets such as 1:1, 4:3, 16:9, and 9:16 work too.
Edits That Follow the Source
With no size set, an edit keeps the aspect ratio of the first reference image, within a 2048 × 2048 area. Set a size or ratio and the edit is delivered at exactly that shape instead.
Prompt Expansion
A short prompt is developed into a complete image description before the render. It is on by default for text-to-image and can be switched off when the prompt should be used exactly as written.
Long Prompts
The prompt can run to 32,000 characters, enough for a full brief with layout notes, exact wording, and a role for each reference image.
A Faster Variant
MiniMax H3 Image Highspeed takes the same inputs and is the quicker of the two, at twice the image rates. Draft on one and finish on the other without changing the prompt.
Where teams reach for MiniMax H3 Image
- Campaign compositions from many referencesPut a model, a product, a location, and props from separate photos into one image, each named by its number in the prompt.
- Product in a new sceneRestage an approved product photo in a different setting and compare the result with the source before using it.
- Bundles and gift setsCombine up to sixteen product shots into a single flat lay or range photo from one instruction.
- Large-format finalsRender at up to 20 megapixels for print, out-of-home, and crops of one master into several placements.
- Unusual shapesGenerate tall or wide formats, up to 3:1 or 1:3, for banners, story placements, and packaging panels.
- Concept images from a promptGenerate mood frames and campaign concepts from text when no reference exists yet.
Frequently asked questions
What teams need to know about creating with MiniMax H3 Image in Masonry
What is MiniMax H3 Image?
MiniMax H3 Image is the image model in MiniMax's H3 family. In Masonry it generates images from a text prompt and edits or combines one to sixteen reference images, through fal's text-to-image and edit endpoints.
How many reference images can MiniMax H3 Image take?
Up to sixteen in one request. Refer to them in the prompt as Image 1, Image 2, and so on, in the order you attached them. A seventeenth image is refused rather than dropped.
What image sizes can it create?
Any width and height from 655,360 to 20,000,000 total pixels, with each edge between 468 and 7,745 pixels and an aspect ratio between 1:3 and 3:1. With no size set, a new image is 2048 × 2048 and an edit follows the shape of its first reference.
What is prompt expansion?
Before rendering, the model develops a short prompt into a complete image description. It is on by default for text-to-image and can be switched off. Edits have no switch for it.
What is the difference between MiniMax H3 Image and H3 Image Highspeed?
Highspeed is the faster variant. It takes the same inputs, including sixteen references and the same sizes, and fal charges twice the image rates for it.
Does it support seeds or negative prompts?
No. fal's endpoints take a prompt, reference images, a size, and an output format (PNG, JPEG, or WebP). There is no seed, negative prompt, or quality setting.
How much does MiniMax H3 Image cost in Masonry?
fal bills it by tokens, so the cost depends on the output size, the number of reference images, and the prompt. Masonry converts that to credits and shows the exact cost before you generate.
Can the output be used commercially?
fal marks the MiniMax H3 Image endpoints for commercial use. Review the current fal and MiniMax terms for your project, particularly for trademarks and real people's likenesses.
How does my team use MiniMax H3 Image in Masonry?
Open a Masonry canvas, pick MiniMax H3 Image from the model selector, and describe the image you need: a product shot, an ad creative, a social post. Masonry generates it, then you refine, edit, and combine MiniMax H3 Image with other models in one workspace.
Is MiniMax H3 Image free to try?
New Masonry accounts get 50 free credits to try GPT Image 2.5 Flare. MiniMax H3 Image is included in every paid plan, and the editor shows the exact credit cost before you generate.
How do I write good prompts for MiniMax H3 Image?
For edits, name each reference by its number and its job: "the bottle in Image 1", "the kitchen in Image 2". Say what must stay the same, such as a label, a face, or a brand color, and what should change. For text-to-image, a short prompt is expanded for you; switch expansion off when you have written the full description yourself and want it followed as written. There is no negative prompt, so describe what you want in the frame instead of listing what to leave out.
Who makes MiniMax H3 Image?
MiniMax H3 Image is built by MiniMax. Inside Masonry it runs alongside 50+ image and video models, so your team can pick the right one for each brief without switching tools.
Start creating with MiniMax H3 Image
Generate, edit, and compare across 50+ models in one workspace.
Guides for MiniMax H3 Image
Prompt walkthroughs and examples from the Masonry blog
Explore more AI models
Compare MiniMax H3 Image with other models teams run in Masonry