Skip to main content

MiniMax's H3-Image, released as open weights and built on its H3 video model, turns a written scene into a still image in five size tiers from 1k to 4k, with an optional seed for comparing one change at a time.

From $0.024/image

Workbench

Input

Text to image
222/5000
Provide up to 5,000 characters describing the desired scene; this required field has no default and no prompt-length fee. at least 1 characters · at most 5000 characters
Pick any of Fifteen options: 1:1, 16:9, 9:16, 4:3, 3:4, 3:2, 2:3, 5:4, 4:5, 2:1, 1:2, 3:1, 1:3, 21:9, 9:21; omitted values use 1:1, without affecting the tier price.
affects price
Generation starts at 1k unless you choose 1.5k, 2k, 3k or 4k; increasing the tier increases the per-image rate.
Output format
JPEGPNGWebP
jpeg is the default encoding; png and webp are available with no change in price.
Seed
Optional seed holds the random source while you compare a single change, without guaranteeing the same output. No numeric bounds or fixed default are specified here, and it adds no cost. no lower than 0 · no higher than 2147483647
Safety checkerOff

SpicyAPI does not run a safety checker on this model. Your prompt and media go to the model as written, and the model itself decides what comes back.

Checking your account before enabling submission and pricing.

Output

Ready for a new run
rooftop-pool-nightExample result

Missing the model you need?

MINIMAX OPEN-WEIGHT IMAGE API

MiniMax H3-Image Open Weights API at a glance

Five pixel tiers and a seed for comparing one change at a time.

Size tiers
5 tiers
Choose 1k, 1.5k, 2k, 3k or 4k, with the rate increasing by tier.
Controlled comparisons
Seed
Use the same random source while adjusting one instruction or setting.
Ordered editing pictures
1–9
The first image is edited and up to eight additional images guide the change.
Default delivery
JPEG
Request PNG or WebP instead without a format charge.
Observed square at 3k
3072²
In our tests the 3k breakfast image measured 3072 x 3072.

Results

What you can make with MiniMax H3-Image Open Weights

  • A man in his thirties steps out of a rooftop pool at night, with water on his shoulders, city lights and blue pool glow behind him.

    Night portrait at 1.5k

    Blue pool light against the skyline

    At 1.5k and 3:4 with seed 7, the rooftop pool example shows a man stepping out of the water at night, with city lights and blue pool glow behind him. It gives the intermediate tier a portrait example alongside the larger square breakfast scene.

    Generated rooftop pool scene at 1.5k, 3:4, seed 7, saved as PNG.

  • An overhead breakfast arrangement contains sourdough, halved figs, a honey jar with a wooden dipper and a linen napkin.

    Square 3k result

    A tabletop scene at the middle-large tier

    The breakfast sample returned a 3072 x 3072 view of bread, figs, honey and linen, with seed 7.

    3k square output, seed 7.

  • A woman with a bicycle stands beside a pale blue vintage coupe on a coastal road in afternoon light.

    Reference roles

    Bring the cyclist into the coastal scene

    The composite kept the road and pale blue coupe from <Picture 1> and placed the cyclist and bicycle from <Picture 2> beside them.

    Two-reference coastal composite at 1k.

  • A woman in a denim jacket stands with her bicycle on a stone bridge amid orange leaves, golden light and river mist.

    Single-source revision

    Exchange overcast weather for autumn light

    One source image became an autumn bridge scene with falling orange leaves and a misty river.

    Autumn edit from one reference, with no aspect ratio supplied.

Content test

Is MiniMax H3-Image Open Weights uncensored? What we tested

Content levels tested on MiniMax H3-Image Open Weights, by mode
LevelText to imageImage editing
Mild
RendersTested Nov 11, 2026
Not tested yet
Moderate
RendersTested Nov 11, 2026
RendersTested Nov 11, 2026
Strong
RendersTested Nov 11, 2026
RendersTested Nov 11, 2026
Very strong
RendersTested Nov 11, 2026
RendersTested Nov 11, 2026
Maximum
RendersTested Nov 11, 2026
Not tested yet
How we test

How each mode behaves

All modes
Sharing a seed makes single-change comparisons useful without ensuring identical images.
All modes
This open-weight interface accepts at most nine editing inputs and does not expose sampling steps, guidance or a negative prompt.
All modes
For requests using these weights, every person you describe must be an adult.
All modes
Downloads of generated files expire after 14 days, when the files are deleted.

Use cases

What to build with the MiniMax H3-Image Open Weights API

  • Night portraits with a defined mood

    Build an editorial scene around a person, an action and a light source. The rooftop pool example returned a man against city lights and blue pool glow at 1.5k, with a 3:4 frame and seed 7.

  • Tabletop arrangements at larger sizes

    Describe the objects and their placement before choosing the output tier. Our 3k square breakfast result measured 3072 x 3072 and showed bread, figs, honey and linen arranged for an overhead view.

  • Composites with separate source roles

    Give the setting and the added subject different picture numbers. In the coastal edit, the road and pale blue coupe remained in the scene while the cyclist and bicycle from another reference appeared beside them.

  • Seasonal changes around a retained subject

    Ask for a new season and lighting while naming the details to preserve. Our bridge edit kept the cyclist and bicycle and added autumn leaves, golden light and mist over the river.

  • Signs with short supplied wording

    Quote a short label or sign and say where it belongs in the scene. The matched-prompt comparison rendered HOT SOUP 3 FRANCS correctly on a wooden sign; longer passages still need careful review.

Prompts

MiniMax H3-Image Open Weights prompt examples you can copy

  • Text to image

    Rooftop after the evening swim

    A man in his thirties stepping out of a rooftop pool at night, water running down his bare chest and shoulders, wet hair pushed back, city lights and blue pool glow behind him, confident look at the camera, photorealistic.

    Why it works. The person, action, light and setting form a complete scene; select 3:4 and 1.5k in the controls for the shape and tier used by the rooftop example.

    Try this prompt
  • Text to image

    Bread, figs and morning linen

    Overhead still life of a rustic breakfast table: sourdough loaf, figs cut in half, a jar of honey with a wooden dipper, linen napkin, morning light, food magazine photography.

    Why it works. Named objects and their positions make this a useful starting point for a square tabletop study; inspect every item rather than assuming the list was followed.

    Try this prompt
  • Text to image

    Camel coat under paper lanterns

    A woman in a camel trench coat walking through a narrow rainy alley in Tokyo at night, paper lanterns and wet cobblestones reflecting orange light, cinematic wide shot, shallow depth of field.

    Why it works. Wardrobe, weather and light explain the visual direction in prose; the rainy alley example provides a related 1.5k widescreen reference.

    Try this prompt
  • Image editing

    Meet beside the coastal coupe

    Keep the coastal road and the pale blue coupe from <Picture 1>. Add the woman and her bicycle from <Picture 2> standing beside the car, chatting with the camera, same afternoon light.

    Why it works. Each numbered source has a specific contribution, which follows the structure of our successful two-picture composite. Use adult reference subjects whose photos you have the right to use; no impersonation.

    Try this prompt
  • Image editing

    Golden leaves around the cyclist

    <Picture 1> is a woman with a bicycle. Keep her face, red curly hair and denim jacket. Change the season to late autumn: orange leaves falling, a misty river under the bridge, golden morning light.

    Why it works. A preservation clause separates the retained subject from the seasonal change; leaving the ratio unset lets the source shape guide the edit. Use a photo you have the right to use, depicting an adult, with no impersonation.

    Try this prompt
  • Text to image

    A cafe sign in the rain

    Create a street photograph of a small cafe after sunset, with a wooden sign beside the doorway bearing the words "HOT SOUP 3 FRANCS" on three clearly separated lines. Warm interior light falls onto wet pavement, a dark green awning shelters the entrance and no people are present.

    Why it works. Short quoted wording with a named location gives you a concrete lettering check; this is a new prompt idea rather than an additional measured sample.

    Try this prompt

Pricing

MiniMax H3-Image Open Weights API pricing

One image starts at $0.024 at 1K and $0.204 at 4K.

Get started

How to use the MiniMax H3-Image Open Weights API in three steps

  1. Step 1

    Start from a scene or a source

    Choose text to image for a written brief, or image editing for a picture you want to revise. For an edit, use photos you have the right to use, with adult subjects and no impersonation.

  2. Step 2

    Describe the frame and reference roles

    Write the setting, subject, wardrobe, light and camera view as connected English prose. In an edit, call the source <Picture 1> and give each later reference its own numbered role.

  3. Step 3

    Set shape, tier and file format

    Choose a resolution from 1k through 4k and select an aspect ratio for generation. Editing follows the source shape when no ratio is supplied; JPEG is the default, with PNG and WebP also available.

Parameter reference and code samples are in the API and Readme tabs at the top of this page.

FAQ

MiniMax H3-Image Open Weights API FAQ

About MiniMax H3-Image Open Weights

Start with a written scene when you need a new composition, or bring a source picture when the subject and setting already matter. The rooftop portrait, square breakfast and cyclist edits show distinct uses of the five-tier controls. Choose the frame and size deliberately, then use a seed to compare one change at a time. Check faces, object shapes and quoted text before accepting a result: our tests include readable short signs and successful composites, but also an object mismatch and toned-down requests. Move to the LoRA sibling when a separate style-weight file is part of your brief.

All MiniMax models21

Related models and next tasks