Veo 3.1 Fast Video Generation

Generate videos using Google Veo 3.1 Fast model. Optimized for speed while maintaining high quality.

POST /vendors/google/v1/veo-3.1-fast/generation

Create Generation Task

cURL

curl --request POST \
  --url https://api.mulerun.com/vendors/google/v1/veo-3.1-fast/generation \
  --header 'Authorization: Bearer <token>' \
  --header 'Content-Type: application/json' \
  --data '
{
  "prompt": "A serene sunset over a calm ocean, with gentle waves lapping against the shore",
  "negative_prompt": "blurry, low quality, pixelated",
  "aspect_ratio": "16:9",
  "resolution": "1080p",
  "duration": 8
}
'

202

{
  "task_info": {
    "id": "3c90c3cc-0d44-4b50-8888-8dd25736052a",
    "created_at": "2023-11-07T05:31:56Z",
    "updated_at": "2023-11-07T05:31:56Z"
  }
}

Beta

This model is currently in public testing. Not everyone has access, and API requests may also be unstable.

Overview

Generate videos with faster processing speed using Google’s Veo 3.1 Fast model. This variant prioritizes speed over maximum quality, making it ideal for rapid prototyping and iteration.

Key Features

• Faster generation: Reduced processing time compared to standard Veo 3.1
• Text-to-Video: Generate videos from descriptive text prompts
• Image-to-Video: Animate a starting image into a video sequence
• Frame interpolation: Generate videos by specifying first and last frames
• Reference images: Use up to 3 reference images to guide style and content
• Audio generation: Natively generates synchronized audio with video

Supported Aspect Ratios and Resolutions

Aspect Ratio	Resolution	Duration Options
16:9	720p, 1080p	4s, 6s, 8s
9:16	720p, 1080p	4s, 6s, 8s

Prompt Writing Tips

For best results, include these elements in your prompt:

• Subject: The main focus (object, person, animal, scenery)
• Action: What the subject is doing (walking, running, turning)
• Style: Creative direction (sci-fi, horror film, film noir, cartoon)
• Camera positioning (optional): aerial view, eye-level, dolly shot
• Composition (optional): wide shot, close-up, single-shot
• Ambiance (optional): blue tones, night, warm tones

Audio Prompting

Veo 3.1 Fast can generate synchronized audio. Include audio cues in your prompt:

• Dialogue: Use quotes for specific speech (e.g., “This must be the key,” he murmured)
• Sound Effects: Explicitly describe sounds (e.g., tires screeching loudly)
• Ambient Noise: Describe the environment’s soundscape (e.g., a faint, eerie hum)

Example Requests

Text-to-Video

{
  "prompt": "A serene sunset over a calm ocean, with gentle waves lapping against the shore",
  "negative_prompt": "blurry, low quality, pixelated",
  "aspect_ratio": "16:9",
  "resolution": "1080p",
  "duration": 8
}

Image-to-Video

{
  "prompt": "The character starts walking forward slowly",
  "image": "data:image/png;base64,iVBORw0KGgoAAAANSUhEUg...",
  "aspect_ratio": "9:16",
  "resolution": "720p",
  "duration": 6
}

With Reference Images

{
  "prompt": "A character walking in a city street at night",
  "reference_images": [
    "data:image/png;base64,iVBORw0KGgoAAAANSUhEUg...",
    "data:image/png;base64,iVBORw0KGgoAAAANSUhEUg..."
  ],
  "aspect_ratio": "16:9",
  "resolution": "720p",
  "duration": 8
}

When using reference images, duration must be set to 8 seconds.

Interpolation (First and Last Frame)

{
  "prompt": "A smooth transition between two scenes",
  "image": "data:image/png;base64,iVBORw0KGgoAAAANSUhEUg...",
  "last_frame": "data:image/png;base64,iVBORw0KGgoAAAANSUhEUg...",
  "aspect_ratio": "16:9",
  "resolution": "720p",
  "duration": 8
}

When to Use Veo 3.1 Fast

• Rapid prototyping: Quickly iterate on prompts and concepts
• Draft generation: Create initial versions before final high-quality renders
• High-volume workflows: When processing many videos in parallel
• Time-sensitive projects: When speed is more important than maximum quality

Authorizations

Authorization · string · header · required

Bearer authentication header of the form Bearer <token>, where <token> is your auth token.

Body

application/json

prompt · string · required

Text description for the video. Supports audio cues.

Use descriptive language including:

• Subject (object, person, animal, scenery)
• Action (what the subject is doing)
• Style (sci-fi, horror film, film noir, cartoon, etc.)
• Camera positioning and motion (optional): aerial view, eye-level, dolly shot
• Composition (optional): wide shot, close-up, single-shot
• Ambiance (optional): blue tones, night, warm tones

Audio Prompting:

• Dialogue: Use quotes for specific speech (e.g., "This must be the key," he murmured)
• Sound Effects: Explicitly describe sounds (e.g., tires screeching loudly)
• Ambient Noise: Describe the environment's soundscape (e.g., a faint, eerie hum)

Maximum string length: 2000

negative_prompt · string

Text describing what not to include in the video.

Do not use instructive language like "no" or "don't". Instead, describe what you don't want to see (e.g., "wall, frame" instead of "No walls").

Maximum string length: 500

image · string | null

Initial image to animate (first frame). Can be a URL or Base64 encoded data.

Format for Base64: data:image/png;base64,{base64_data}

Supported formats: JPEG, JPG, PNG, BMP, WEBP Max file size: 20MB

last_frame · string | null

Final image for interpolation video. Must be used in combination with the image parameter.

Format for Base64: data:image/png;base64,{base64_data}

Supported formats: JPEG, JPG, PNG, BMP, WEBP Max file size: 20MB

reference_images · string[] | null

Up to 3 images to be used as style and content references. Provide images of a person, character, or product to preserve the subject's appearance in the output video.

Each item can be a URL or Base64 encoded data. Format for Base64: data:image/png;base64,{base64_data}

Note: When using reference images, duration must be 8 seconds.

Maximum array length: 3

aspect_ratio · enum<string> · default: 16:9

Video aspect ratio (width:height).

Available options: 16:9, 9:16

resolution · enum<string> · default: 720p

Video resolution.

Note: 1080p and 4k only support 8 second duration.

Available options: 720p, 1080p, 4k

duration · enum<integer> · default: 8

Length of the generated video in seconds.

Note: Must be 8 when using reference images, 1080p, or 4k resolution.

Available options: 4, 6, 8

Response

202 - application/json

Accepted - Task created successfully

task_info · object

Show child attributes