Saltar al contenido principal
Model logo

API de MiniMax H3 Max

  • Texto a vídeo
  • Imagen a vídeo
  • Referencia a vídeo

MiniMax H3 Max forma parte de la familia de modelos MiniMax HailuoVer todos los modelos MiniMax Hailuo

Sube imágenes/vídeos arriba y luego escribe @ en tu prompt para referenciarlos

Toca para subirPNG, JPG, WEBP, GIF · Máx. 100 MB
Toca para subirPNG, JPG, WEBP, GIF · Máx. 100 MB
Toca para subirPNG, JPG, WEBP, GIF · Máx. 100 MB
Toca para subirMP4, WEBM, MOV · Máx. 100 MB
Toca para subirMP3, WAV, FLAC, OGG · Máx. 100 MB
Salida

Tu video generado aparecerá aquí

Actualizado

What is the MiniMax H3 Max API?

The MiniMax H3 Max API is live on UnificAlly as hailuo/minimax-h3-max, from $0.045 per second of video. MiniMax H3 Max is the 480p and 768p variant of MiniMax H3 with the full H3 control set. It renders video at 24 fps with native audio, runs 5 to 15 seconds per generation, and takes text alone, a start frame with an optional end frame, 1 to 9 reference images, up to 3 reference videos, and up to 3 reference audio clips alongside images or videos. A seed and a prompt expansion level (disabled, balanced, quality) are optional on every request. For a faster, text-and-frames-only version, see MiniMax H3 Max Turbo.

MiniMax H3 Max API specs

Model IDhailuo/minimax-h3-max
EndpointPOST https://api.unifically.com/v1/tasks
Request typesText-to-video, start and end frames, reference images, videos and audio
Duration5 to 15 seconds, any whole second
Resolution480p, 768p
Aspect ratio21:9, 16:9, 4:3, 1:1, 3:4, 9:16
AudioNative audio in the output
ReferencesUp to 9 images, 3 videos, 3 audio clips. Videos and audio together at most 15 seconds
Rate at 480p$0.045/sec
Rate at 768p$0.072/sec

Billing is per second of output, with no subscription. The first 2 reference images are included, each extra image and each second of reference video is billed, and reference audio is not charged. The MiniMax pricing guide covers the full rate table.

How to use the MiniMax H3 Max API

  1. Create an API key on the API keys page.
  2. Send a POST request to /v1/tasks with the model ID and an input object.
  3. Read the task_id from the response, then check the task until it is finished, or pass a callback URL and wait for the result.
curl -X POST https://api.unifically.com/v1/tasks \
  -H "Content-Type: application/json" \
  -H "Authorization: Bearer YOUR_API_KEY" \
  -d '{
    "model": "hailuo/minimax-h3-max",
    "input": {
      "prompt": "A slow dolly shot across a rain-soaked neon street, reflections rippling in the puddles",
      "aspect_ratio": "16:9",
      "resolution": "768p",
      "duration": 5,
      "prompt_expansion": "balanced"
    }
  }'

The response returns the task ID straight away:

{
  "code": 200,
  "success": true,
  "data": {
    "task_id": "abc123def456",
    "status": "processing"
  }
}

Poll the task until its status is completed or failed. The finished task carries the video URL.

curl https://api.unifically.com/v1/tasks/{task_id} \
  -H "Authorization: Bearer YOUR_API_KEY"

MiniMax H3 Max API parameters

ParameterTypeDefaultWhat it does
promptstringrequiredVideo prompt, 1 to 2000 characters. With reference images, use @Image1 through @Image9 to name them
image_urlsstring[]none1 to 9 public reference image URLs. The first 2 are included, each extra image is billed. Cannot be combined with frames
video_urlsstring[]none1 to 3 public reference video URLs, 15 seconds in total. Billed per second of uploaded video. Cannot be combined with frames
audio_urlsstring[]none1 to 3 public reference audio URLs. Reference videos and audio together at most 15 seconds. Needs image_urls or video_urls. Cannot be combined with frames
start_frame_urlstringnonePublic start-frame image URL. Cannot be combined with image_urls or video_urls
end_frame_urlstringnonePublic end-frame image URL. Requires start_frame_url
aspect_ratiostring16:921:9, 16:9, 4:3, 1:1, 3:4, or 9:16. Omit it with frames: the frame sets the shape
resolutionstring768p480p or 768p
durationinteger5Output length in whole seconds, 5 to 15
seedintegerrandom0 to 4294967295 for repeatable output
prompt_expansionstringbalancedHow much the model rewrites the prompt before generating: disabled, balanced, or quality
nsfw_checkbooleanfalseCheck the prompt and input images for NSFW content before generating

The MiniMax H3 Max API documentation has the full reference for the whole MiniMax H3 family.

Reference image request

Name each reference image in the prompt in array order, @Image1 through @Image9.

curl -X POST https://api.unifically.com/v1/tasks \
  -H "Content-Type: application/json" \
  -H "Authorization: Bearer YOUR_API_KEY" \
  -d '{
    "model": "hailuo/minimax-h3-max",
    "input": {
      "prompt": "@Image1 walks through the location shown in @Image2",
      "image_urls": [
        "https://example.com/character.jpg",
        "https://example.com/location.jpg"
      ],
      "aspect_ratio": "16:9",
      "resolution": "768p",
      "duration": 6
    }
  }'

Key features of MiniMax H3 Max

The full H3 control set at 480p or 768p

Reference images, reference videos, reference audio, start and end frames, a seed and a prompt expansion level all work the same way as on H3, so a request tuned here moves to H3 unchanged when 2K is needed.

Reference-to-video

Pass 1 to 9 reference images and name them in the prompt as @Image1 through @Image9. The first 2 images are included in the per-second price and each extra image is billed. Reference images cannot be combined with start/end frames in one request.

Reference videos and audio

Up to 3 reference videos in video_urls and up to 3 audio clips in audio_urls, 15 seconds in total across both. Each second of uploaded video is billed; audio is free and needs reference images or videos in the same request.

Seed and prompt expansion

A seed repeats a result with the same inputs. prompt_expansion sets how much the model rewrites the prompt: disabled sends it as written, balanced is the default, quality rewrites more.

Native audio in the same pass

Output arrives with an audio track generated alongside the picture, no separate audio step.

What MiniMax H3 Max is best for

  • Character-locked series

    Reference images keep the same character across every video in a campaign, at 768p with sound.

  • Character-in-location shots

    Pass the character as @Image1 and the setting as @Image2, and the prompt places one inside the other.

  • Repeatable renders

    Fix the seed and set prompt_expansion to disabled to get the same result from the same request.

  • Start-and-end frame spots

    Pin the open and the close as key art and let the model animate the move between them.

MiniMax H3 Max use cases

Run a product campaign where the same mascot appears in ten videos by sending the character sheet as reference images. Put a product into a new scene by passing the product and the location as two reference images. Build a vertical 9:16 spot with sound for social feeds in one call. Pass a reference video and an audio clip to carry their motion and sound into a new shot. For pinned compositions, pass a start frame, optionally with an end frame, and the output follows the input frame's ratio.

MiniMax H3 Max limitations

H3 Max caps at 768p; for 2K output use MiniMax H3. The shortest generation is 5 seconds, not 4. Only the first 2 reference images are included in the price; MiniMax H3 Fast charges nothing per image. Reference images and frame images are mutually exclusive in one request, an end frame needs a start frame, and reference audio needs reference images or videos. Duration is capped at 15 seconds per generation and reference videos plus audio at 15 seconds in total.

Preguntas frecuentes

Preguntas frecuentes sobre MiniMax H3 Max

Resolution, minimum duration and reference-image billing. H3 Max outputs 480p or 768p and runs 5 to 15 seconds; H3 outputs 768p or 2K and runs 4 to 15 seconds. Both take a prompt, a start frame with an optional end frame, up to 9 reference images, up to 3 reference videos, up to 3 reference audio clips, a seed and a prompt expansion level, with native audio. H3 Max includes the first 2 reference images in the price; H3 includes the first 5.

Start with a key from the API keys page on your UnificAlly account and send it as Authorization: Bearer <key> to https://api.unifically.com/v1/tasks with the model hailuo/minimax-h3-max. The first 2 reference images are included; extra images and reference video seconds are billed on top.

Along with hailuo/minimax-h3-max, send a prompt, resolution (480p or 768p), a duration of 5 to 15 seconds, aspect_ratio and optional prompt_expansion and seed. Use frames or references (up to 9 image_urls, 3 video_urls, 3 audio_urls), not both, and name images @Image1 to @Image9 in the prompt. Poll the returned task_id until completed or failed, or set a callback URL, to read the video URL.

$0.045/sec at 480p and $0.072/sec at 768p, billed per second of output. The first 2 reference images are included and each extra image is billed; a reference video is billed per second of uploaded video. Reference audio is free.

480p and 768p through the resolution parameter. The default is 768p. aspect_ratio controls the output shape when no start or end frame is supplied; pick a fixed ratio (the auto option exists only on MiniMax H3 and H3 Lite).

Yes. Pass up to 3 reference videos in video_urls and up to 3 audio clips in audio_urls, 15 seconds in total across both. Audio needs reference images or videos in the same request. Each second of uploaded video is billed; audio is free.

No. image_urls cannot be combined with start_frame_url or end_frame_url in one request. Use frames for pinned compositions and reference images for subject or style control. An end frame also needs a start frame.

Yes. seed makes a result repeatable with the same inputs, and prompt_expansion (disabled, balanced, quality) sets how much the model rewrites the prompt before generating.

Any whole second from 5 to 15 per generation. The default is 5; there is no 4-second option on H3 Max.

Guías y comparativas de MiniMax H3 Max

Consulta precios, límites y resultados probados antes de usar este modelo.