MiniMax H3 Max Spicy Reference to Video API: Image-to-Video

Generate video with MiniMax H3 Max Spicy on the Muapi API. Relaxed-moderation tier of MiniMax H3 Max at 480p/768p. Pay per generation, no subscription.

Interactive model controls

MiniMax H3 Max Spicy Reference to Video is the relaxed-moderation tier of MiniMax H3 Max, generating video with native audio at 480p or 768p through the Muapi API.

📝

Overview

About this model

MiniMax H3 Max Spicy Reference to Video is the relaxed-moderation tier of MiniMax H3 Max. It generates a video clip guided by reference images, videos and audio with native audio at 480p or 768p, 5-15 seconds per clip, using the same inputs as the standard model, so existing integrations can switch endpoints with no payload changes. The Spicy tier accepts bolder, more expressive creative briefs than the standard model, and is priced at $0.0575/second at 480p and $0.092/second at 768p - 15% above the standard MiniMax H3 Max rates. It follows Muapi's standard asynchronous request, poll and webhook pattern. 1080p is available only on the standard model, and custom seeds and prompt-expansion modes are not supported on this tier.

1Creative storytelling: Produce bold, expressive scenes that standard moderation tiers may reject.
2Character content: Build consistent character clips for adult-oriented creative projects.
3Ad and social testing: Generate many variants quickly to test hooks and visuals.
4Previsualization: Rough out camera moves and shot composition before a production shoot.
💰

Pricing & Value

Cost analysis

muapiapp$0.0575/second at 480p and $0.092/second at 768p

15% above standard MiniMax H3 Max pricing; billed by resolution and duration.

Fal.aiNot available

No Spicy tier.

ReplicateNot available

Not available.

** Competitor pricing is estimated based on similar model architectures and usage tiers.

⚙️

Technical Details

Configuration schema

Promptstring

Describe the video scene and how the references should be used (e.g. 'Image 1', 'Video 1').

Default ValueUse the references to create a cinematic product reveal with a slow camera orbit.
Reference Imagesarray

Reference image URLs (up to 9). Provide at least one image or video reference. Max 12 combined reference files.

Default Valueundefined
Reference Videosarray

Reference video URLs (up to 3; 2-15s each, 15s combined max). Provide at least one image or video reference.

Default Valueundefined
Reference Audioarray

Optional reference audio URLs (up to 3; 2-15s each, 15s combined max). Audio cannot be used alone.

Default Valueundefined
Aspect RatioEnum (7 options)

-

Default Valueadaptive
ResolutionEnum (2 options)

-

Default Value768p
DurationEnum (11 options)

Output duration in seconds.

Default Value5
📖

Implementation Guide

Developer documentation

How to Use MiniMax H3 Max Spicy Reference to Video

  1. Add references: Provide at least one entry in reference_images or reference_videos; optional reference_audios guide sound (12 reference files combined at most).

  2. Describe the scene: Write prompt describing how the references should appear and move.

  3. Pick resolution and duration: Set resolution to 480p or 768p and duration to any value from 5 to 15 seconds.

  4. Submit and retrieve: POST the request, then poll /predictions/{request_id}/result or supply a webhook_url to receive the finished video URL.

curl -X POST https://api.muapi.ai/api/v1/minimax-h3-max-spicy-reference-to-video \
  -H "x-api-key: YOUR_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "prompt": "The character from the reference image walks through a rainy street at night.",
    "reference_images": ["https://example.com/character.jpg"],
    "resolution": "768p",
    "duration": 5
  }'
❓

Common Questions

Frequently asked

What is MiniMax H3 Max Spicy Reference to Video?

It is the relaxed-moderation tier of MiniMax H3 Max. It generates a video clip guided by reference images, videos and audio with native audio and accepts bolder creative prompts than the standard model.

What resolutions and durations are supported?

Resolution can be 480p or 768p, and duration can be any value from 5 to 15 seconds. 1080p is only available on the standard model.

How much does it cost?

$0.0575/second at 480p and $0.092/second at 768p, which is 15% above the standard MiniMax H3 Max rates. The exact price for a request is calculated from its resolution and duration, plus reference charges.

How is it different from MiniMax H3 Max?

It uses the same inputs, but runs on a relaxed-moderation tier, is limited to 480p and 768p, and does not support custom `seed` or `prompt_expansion_mode` values.

How do I get the finished video?

Requests are asynchronous. Poll `/predictions/{request_id}/result` with the returned `request_id`, or pass a `webhook_url` to be notified when the video is ready.

What happens if a generation fails?

Failed generations are refunded automatically.