MiniMax

Video-01 Base

MiniMax's Video-01 Base model provides foundational video generation with cinematic motion control, supporting text-to-video, image-to-video, and reference-to-video workflows.

Model AIR ID: minimax:1@1.

Supported workflows: Text-to-video, image-to-video, reference-to-video.

Technical specifications:

• Positive prompt: 2-2000 characters (optional when using frameImages).
• Supported dimensions: 1366×768 (768p).
• Frame rate: 25 FPS.
• Duration: 6 seconds.
• Frame images: Supports first frame for frameImages.
• Reference images: Supports referenceImages with 1 image.
• Input image requirements: Width and height between 300-2048 pixels, 20MB file size limit.

Provider-specific parameters supported: promptOptimizer.

{
  "taskType": "videoInference",
  "taskUUID": "24cd5dff-cb81-4db5-8506-b72a9425f9d1",
  "model": "minimax:1@1",
  "positivePrompt": "A cinematic tracking shot following a car driving through a mountain road at sunset",
  "duration": 6,
  "width": 1366,
  "height": 768,
  "providerSettings": {
    "minimax": {
      "promptOptimizer": true
    }
  }
}

{
  "taskType": "videoInference",
  "taskUUID": "b8c4d952-7f27-4a6e-bc9a-83f01d1c6d59",
  "model": "minimax:1@1",
  "frameImages": [
    {
      "inputImage": "c64351d5-4c59-42f7-95e1-eace013eddab",
      "frame": "first"
    }
  ],
  "duration": 6,
  "width": 1366,
  "height": 768
}

Video-01 Live

MiniMax's Video-01 Live model specializes in bringing static illustrations to life with vivid, expressive character animations optimized for 2D artwork and avatar creation.

Model AIR ID: minimax:2@3.

Supported workflows: Image-to-video.

Technical specifications:

• Positive prompt: 2-2000 characters (optional).
• Supported dimensions: 1366×768 (768p).
• Frame rate: 25 FPS.
• Duration: 6 seconds.
• Frame images: Supports first frame for frameImages.
• Input image requirements: Width and height between 300-2048 pixels, 20MB file size limit.

Provider-specific parameters supported: promptOptimizer.

{
  "taskType": "videoInference",
  "taskUUID": "f3a2b8c9-1e47-4d3a-9b2f-8c7e6d5a4b3c",
  "model": "minimax:2@3",
  "frameImages": [
    {
      "inputImage": "c64351d5-4c59-42f7-95e1-eace013eddab",
      "frame": "first"
    }
  ],
  "duration": 6,
  "width": 1366,
  "height": 768
}

Video-01 Director

MiniMax's Video-01 Director model provides enhanced cinematic control with precise filmmaking capabilities, offering reduced randomness and granular camera movement control for professional storytelling.

Model AIR ID: minimax:2@1.

Supported workflows: Text-to-video, image-to-video.

Technical specifications:

• Positive prompt: 2-2000 characters (optional when using frameImages).
• Supported dimensions: 1366×768 (768p).
• Frame rate: 25 FPS.
• Duration: 6 seconds.
• Frame images: Supports first frame for frameImages.
• Input image requirements: Width and height between 300-2048 pixels, 20MB file size limit.

Provider-specific parameters supported: promptOptimizer.

{
  "taskType": "videoInference",
  "taskUUID": "24cd5dff-cb81-4db5-8506-b72a9425f9d1",
  "model": "minimax:2@1",
  "positivePrompt": "Professional push-in shot of a coffee cup on a wooden table, shallow depth of field, warm lighting",
  "duration": 6,
  "width": 1366,
  "height": 768,
  "providerSettings": {
    "minimax": {
      "promptOptimizer": true
    }
  }
}

{
  "taskType": "videoInference",
  "taskUUID": "b8c4d952-7f27-4a6e-bc9a-83f01d1c6d59",
  "model": "minimax:2@1",
  "frameImages": [
    {
      "inputImage": "c64351d5-4c59-42f7-95e1-eace013eddab",
      "frame": "first"
    }
  ],
  "positivePrompt": "Slow tilt shot revealing the scene with cinematic precision",
  "duration": 6,
  "width": 1366,
  "height": 768
}

Hailuo 02

MiniMax's Hailuo 02 model represents next-generation video generation technology with advanced physics simulation, professional camera controls, and support for multiple resolutions and durations.

Model AIR ID: minimax:3@1.

Supported workflows: Text-to-video.

Technical specifications:

• Positive prompt: 2-2000 characters.
• Supported dimensions: 1366×768 (16:9), 1920×1080 (16:9).
• Frame rate: 25 FPS.
• Duration: 6 or 10 seconds (1366×768, default: 6), 6 seconds (1920×1080).
• Frame images: Supports first frame for frameImages.

Provider-specific parameters supported: promptOptimizer.

{
  "taskType": "videoInference",
  "taskUUID": "24cd5dff-cb81-4db5-8506-b72a9425f9d1",
  "model": "minimax:3@1",
  "positivePrompt": "Realistic water physics simulation showing a stone dropping into a calm pond with ripple effects",
  "duration": 10,
  "width": 1366,
  "height": 768,
  "providerSettings": {
    "minimax": {
      "promptOptimizer": true
    }
  }
}

{
  "taskType": "videoInference",
  "taskUUID": "e4d3c2b1-5a6f-4c8e-b2d7-1f0e9d8c7b6a",
  "model": "minimax:3@1",
  "positivePrompt": "Professional tracking shot of a character walking through a bustling marketplace with dynamic lighting",
  "duration": 6,
  "width": 1920,
  "height": 1080,
  "providerSettings": {
    "minimax": {
      "promptOptimizer": true
    }
  }
}

Hailuo 2.3

Built for cinematic storytelling and expressive motion, MiniMax Hailuo 2.3 transforms text or images into visually rich videos with lifelike human physics and precise prompt adherence. It delivers fluid movement and emotional realism with consistent scene composition, ideal for creators seeking artistry and technical polish.

Model AIR ID: minimax:4@1.

Supported workflows: Text-to-video, image-to-video.

Technical specifications:

• Positive prompt: 2-2000 characters (optional when using frameImages).
• Supported dimensions: 1366×768 (16:9), 1920×1080 (16:9).
• Frame rate: 25 FPS.
• Duration: 6 or 10 seconds (1366×768, default: 6), 6 seconds (1920×1080).
• Frame images: Supports first frame for frameImages.
• Input image requirements: Width and height between 300-2048 pixels, 20MB file size limit.

Provider-specific settings:

Parameters supported: promptOptimizer.

{
  "taskType": "videoInference",
  "taskUUID": "24cd5dff-cb81-4db5-8506-b72a9425f9d2",
  "model": "minimax:4@1",
  "positivePrompt": "Cinematic aerial shot of a coastal cliff at sunset with crashing waves",
  "duration": 10,
  "width": 1366,
  "height": 768,
  "providerSettings": {
    "minimax": {
      "promptOptimizer": true
    }
  }
}

{
  "taskType": "videoInference",
  "taskUUID": "b8c4d952-7f27-4a6e-bc9a-83f01d1c6d60",
  "model": "minimax:4@1",
  "frameImages": [
    {
      "inputImage": "c64351d5-4c59-42f7-95e1-eace013eddab",
      "frame": "first"
    }
  ],
  "positivePrompt": "Bring the scene to life with natural movement and atmospheric depth",
  "duration": 10,
  "width": 1366,
  "height": 768
}

{
  "taskType": "videoInference",
  "taskUUID": "e4d3c2b1-5a6f-4c8e-b2d7-1f0e9d8c7b6b",
  "model": "minimax:4@1",
  "frameImages": [
    {
      "inputImage": "c64351d5-4c59-42f7-95e1-eace013eddab",
      "frame": "first"
    }
  ],
  "positivePrompt": "Cinematic reveal with smooth camera movement and realistic lighting",
  "duration": 6,
  "width": 1920,
  "height": 1080
}

Hailuo 2.3 Fast

Accelerated image-to-video version of Hailuo 2.3 built for faster turnaround. Delivers short cinematic clips with consistent motion and high visual clarity.

Model AIR ID: minimax:4@2.

Supported workflows: Image-to-video.

Technical specifications:

• Positive prompt: 2-2000 characters (optional).
• Supported dimensions: 1366×768 (16:9), 1920×1080 (16:9).
• Frame rate: 25 FPS.
• Duration: 6 or 10 seconds (1366×768, default: 6), 6 seconds (1920×1080).
• Frame images: Supports first frame for frameImages.
• Input image requirements: Width and height between 300-2048 pixels, 20MB file size limit.

Provider-specific settings:

Parameters supported: promptOptimizer.

{
  "taskType": "videoInference",
  "taskUUID": "b8c4d952-7f27-4a6e-bc9a-83f01d1c6d61",
  "model": "minimax:4@2",
  "frameImages": [
    {
      "inputImage": "c64351d5-4c59-42f7-95e1-eace013eddab",
      "frame": "first"
    }
  ],
  "positivePrompt": "Add subtle movement and atmosphere to the scene",
  "duration": 6,
  "width": 1366,
  "height": 768
}

{
  "taskType": "videoInference",
  "taskUUID": "e4d3c2b1-5a6f-4c8e-b2d7-1f0e9d8c7b6c",
  "model": "minimax:4@2",
  "frameImages": [
    {
      "inputImage": "c64351d5-4c59-42f7-95e1-eace013eddab",
      "frame": "first"
    }
  ],
  "positivePrompt": "Smooth cinematic motion with high clarity",
  "duration": 6,
  "width": 1920,
  "height": 1080,
  "providerSettings": {
    "minimax": {
      "promptOptimizer": true
    }
  }
}

MiniMax Speech 2.8

MiniMax Speech 2.8 is an advanced text-to-speech model that transforms text into natural, expressive audio across multiple languages. This model delivers broadcast-ready speech with rich prosody, emotional control, and an extensive voice library, making it ideal for voiceovers, narration, accessibility tools, and interactive voice applications.

Model AIR ID: minimax:speech@2.8.

Supported workflows: Text-to-speech.

Technical specifications:

• Text input: Up to 50,000 characters via speech.text.
• Voice library: 332 voices across multiple languages and styles via speech.voice (see table below).
• Speech controls:
  • speech.voice (string, required): Voice identifier from voice library.
  • speech.speed (number): 0.5-2.0, default 1.0.
  • speech.volume (number): 1-10, default 1.
  • speech.pitch (number): -12 to +12, default 0.
  • speech.emotion (string): happy, sad, angry, fearful, disgusted, surprised, calm, fluent, whisper.
  • speech.tone (array): Pronunciation dictionary for tone mapping, e.g., ["omg/oh my god"].
• Generation settings:
  • settings.turbo (boolean): Enable faster, lower-cost generation with reduced quality.
  • settings.languageBoost (string, default: auto): Language detection. Options: zh, yue, en, ar, ru, es, fr, pt, de, tr, nl, uk, vi, id, ja, it, ko, th, pl, ro, el, cs, fi, hi, bg, da, he, ms, fa, sk, sv, hr, fil, hu, no, sl, ca, nn, ta, af, auto.
• Audio output settings:
  • audioSettings.sampleRate (integer, default: 32000): 8000, 16000, 22050, 24000, 32000, 44100 Hz.
  • audioSettings.bitrate (integer, default: 128000): 32, 64, 128, 256 kbps (MP3 only).
  • audioSettings.channels (integer, default: 2): 1 (mono) or 2 (stereo).
• Voice modification:
  • settings.voiceModify.pitch (number, default: 0): -100 to +100.
  • settings.voiceModify.intensity (number, default: 0): -100 to +100.
  • settings.voiceModify.timbre (number, default: 0): -100 to +100.
  • settings.voiceModify.soundEffects (string): spacious_echo, auditorium_echo, lofi_telephone, robotic.

{
  "taskType": "audioInference",
  "taskUUID": "f47ac10b-58cc-4372-a567-0e02b2c3d490",
  "model": "minimax:speech@2.8",
  "speech": {
    "text": "Welcome to our podcast. Today we'll explore the fascinating world of artificial intelligence.",
    "voice": "English_expressive_narrator",
    "speed": 1.0
  },
  "audioSettings": {
    "sampleRate": 32000,
    "bitrate": 128
  }
}

{
  "taskType": "audioInference",
  "taskUUID": "6ba7b833-9dad-11d1-80b4-00c04fd430c8",
  "model": "minimax:speech@2.8",
  "speech": {
    "text": "This is absolutely amazing! I can't believe what just happened!",
    "voice": "English_radiant_girl",
    "emotion": "happy",
    "speed": 1.2,
    "pitch": 2
  },
  "audioSettings": {
    "sampleRate": 44100,
    "bitrate": 256,
    "channels": 2
  }
}

{
  "taskType": "audioInference",
  "taskUUID": "550e8400-e29b-41d4-a716-446655440014",
  "model": "minimax:speech@2.8",
  "speech": {
    "text": "Broadcasting from deep space. All systems operational.",
    "voice": "English_magnetic_voiced_man",
    "volume": 8
  },
  "settings": {
    "voiceModify": {
      "pitch": -30,
      "intensity": 20,
      "soundEffects": "spacious_echo"
    }
  },
  "audioSettings": {
    "sampleRate": 32000
  }
}

{
  "taskType": "audioInference",
  "taskUUID": "a770f077-f413-47de-9dac-be0b26a35da7",
  "model": "minimax:speech@2.8",
  "speech": {
    "text": "Bonjour! Comment allez-vous aujourd'hui?",
    "voice": "French_narrator"
  },
  "settings": {
    "languageBoost": "auto"
  },
  "audioSettings": {
    "sampleRate": 24000,
    "channels": 1
  }
}