> ## Documentation Index
> Fetch the complete documentation index at: https://docs.argil.ai/llms.txt
> Use this file to discover all available pages before exploring further.

# MiniMax H3 Max

> Generate a 5–15 s video with fal’s MiniMax H3 Max from text, from a start (and end) frame, or guided by up to 12 image, video and audio references. 480P/768P, 1080P in reference mode.

Model id `minimax-h3-max` · MiniMax · video · 3 modes: `text-to-video`, `image-to-video`, `reference-to-video`.

> [!NOTE]
>
> This page reflects the current catalog. Inputs, allowed values, limits and prices can change as the model evolves; the field names and their meaning do not. For the exact schema at any time, your agent calls [`get_model`](https://docs.argil.ai/v2/mcp/tools) with the mode.

## Text To Video

Mode `text-to-video`. Text to video.

### Input

| Field | Type | Required | Allowed | Default | Description |
| - | - | - | - | - | - |
| `prompt` | string | Yes | up to 5,000 characters | | What the video should show: subject, action, camera and style. |
| `duration` | integer | Yes | 5 to 15 | | Video duration in seconds. Integer 5–15. |
| `aspect_ratio` | string | | `21:9`, `16:9`, `4:3`, `1:1`, `3:4`, `9:16` | | Aspect ratio. Defaults to 16:9; frames mode follows the start frame. |
| `resolution` | string | | `480P`, `768P` | | 480P (default, fastest) or 768P. |

### Pricing

\~2.42 credits/s at 480P (fastest); 768P is \$0.08/s.

Fixed price: charged as estimated.

For the exact price of a request, use [`estimate_generation`](https://docs.argil.ai/v2/mcp/tools).

### Example

The arguments of [`generate_media`](https://docs.argil.ai/v2/mcp/tools#generate_media):

```json theme={null}
{
  "model": "minimax-h3-max",
  "mode": "text-to-video",
  "input": {
    "prompt": "A ginger cat walks into a sunlit kitchen, handheld camera",
    "duration": 5,
    "aspect_ratio": "21:9",
    "resolution": "480P"
  }
}
```

## Image To Video

Mode `image-to-video`. Animate a start frame, optionally to an end frame.

### Input

| Field | Type | Required | Allowed | Default | Description |
| - | - | - | - | - | - |
| `prompt` | string | Yes | up to 5,000 characters | | What the video should show: subject, action, camera and style. |
| `duration` | integer | Yes | 5 to 15 | | Video duration in seconds. Integer 5–15. |
| `aspect_ratio` | string | | `21:9`, `16:9`, `4:3`, `1:1`, `3:4`, `9:16` | | Aspect ratio. Defaults to 16:9; frames mode follows the start frame. |
| `resolution` | string | | `480P`, `768P` | | 480P (default, fastest) or 768P. |
| `first_frame` | [File](https://docs.argil.ai/v2/mcp/how-it-works.md#your-files) | | | | The image to use as the first frame of the video. |
| `last_frame` | [File](https://docs.argil.ai/v2/mcp/how-it-works.md#your-files) | | | | Optional last frame: the video goes from the first to it. |

### Pricing

\~2.42 credits/s at 480P (fastest); 768P is \$0.08/s.

Fixed price: charged as estimated.

For the exact price of a request, use [`estimate_generation`](https://docs.argil.ai/v2/mcp/tools).

### Example

The arguments of [`generate_media`](https://docs.argil.ai/v2/mcp/tools#generate_media):

```json theme={null}
{
  "model": "minimax-h3-max",
  "mode": "image-to-video",
  "input": {
    "prompt": "A ginger cat walks into a sunlit kitchen, handheld camera",
    "duration": 5,
    "aspect_ratio": "21:9",
    "resolution": "480P"
  }
}
```

## Reference To Video

Mode `reference-to-video`. A video guided by up to 12 image, video and audio references.

### Input

| Field | Type | Required | Allowed | Default | Description |
| - | - | - | - | - | - |
| `prompt` | string | Yes | up to 5,000 characters | | Refer to assets as Image 1, Video 1, Audio 1, in input order. |
| `duration` | integer | Yes | 5 to 15 | | Video duration in seconds. Integer 5–15. |
| `aspect_ratio` | string | | `adaptive`, `21:9`, `16:9`, `4:3`, `1:1`, `3:4`, `9:16` | | Aspect ratio; adaptive follows the references. |
| `resolution` | string | | `480P`, `768P`, `1080P` | | 480P (default), 768P, or 1080P (refined from 768P). |
| `seed` | integer | | | | Seed. Random when omitted. |
| `prompt_expansion_mode` | string | | `balanced`, `quality` | | How the prompt is expanded. balanced by default. |
| `references` | [File](https://docs.argil.ai/v2/mcp/how-it-works.md#your-files) list | | 0 to 36 items | | References the output must follow. The kind of each one (image, video, audio) is read from the file; per-kind limits are in the mode’s `capabilities.references`. |

**References:** up to 36: 12 images, 12 videos of 2 to 15 s (15 s in total), 12 audio files (15 s in total). Assets, URLs and elements are accepted.

### Pricing

2.42/3.87/7.74 credits/s at 480P/768P/1080P; 480P default, plus estimated reference charges.

Fixed price: charged as estimated.

For the exact price of a request, use [`estimate_generation`](https://docs.argil.ai/v2/mcp/tools).

### Example

The arguments of [`generate_media`](https://docs.argil.ai/v2/mcp/tools#generate_media):

```json theme={null}
{
  "model": "minimax-h3-max",
  "mode": "reference-to-video",
  "input": {
    "prompt": "A ginger cat walks into a sunlit kitchen, handheld camera, keeping the look of the reference",
    "duration": 5,
    "aspect_ratio": "adaptive",
    "resolution": "480P",
    "references": [
      "asset_91a3c7e5b2d84f6a8e0c1b9d7f3a5e24"
    ]
  }
}
```


This documentation is built and hosted on [Mintlify](https://mintlify.com), a developer documentation platform.