Skip to main content
RunwayMediumUltra

Runway Gen-4 Image

Runway's Gen-4 Image model with references: combine up to 3 reference images with @tag mentions in your prompt to keep characters, objects, and locations consistent across every angle and scene, at 720p or 1080p.

190 credits
per image at 1080p (default) โ€” 720p 120 credits, 1080p 190 credits
Up to 3 reference images for consistent characters, objects, and locations
@tag_name mentions in the prompt bind each reference to a role
720p and 1080p output resolutions with differential pricing
6 aspect ratios from 21:9 cinematic to 9:16 vertical
Seed support for reproducible generation

Run it right now

Test this model instantly in the Console Playground โ€” no code required

Sign in to try

Use with AI Assistant

Copy usage instructions for Claude, ChatGPT, or other AI

Quick Start

curl -X POST "https://api.core.today/v1/predictions" \
  -H "Content-Type: application/json" \
  -H "X-API-Key: cdt_your_api_key" \
  -d '{
  "model": "runwayml/gen4-image",
  "input": {
    "prompt": "a close up portrait of @woman and @man standing in @park, hands in pockets, looking cool. She is wearing her pink sweater and bangles.",
    "resolution": "1080p",
    "aspect_ratio": "4:3",
    "reference_tags": [
      "park",
      "woman",
      "man"
    ],
    "reference_images": [
      "https://example.com/park.jpg",
      "https://example.com/woman.jpg",
      "https://example.com/man.jpg"
    ]
  }
}'

Parameters

ParameterTypeRequiredDefaultDescription
promptstringYes-Text prompt for image generation. Reference your images using @tag_name
seedintegerNo-Random seed. Set for reproducible generation
resolutionstringNo1080pImage resolution โ€” 720p bills 150 credits, 1080p bills 240 credits
720p1080p
aspect_ratiostringNo16:9Image aspect ratio
16:99:164:33:41:121:9
reference_tagsarrayNo[]An optional tag for each reference image, referenced in your prompt as @tag_name. Alphanumeric, starts with a letter, 3-15 characters
reference_imagesarrayNo[]Up to 3 reference images. Images must be between 0.5 and 2 aspect ratio

How to Provide File Input

There are 3 ways to provide files for the reference_images parameter:

Recommended

Direct Upload (Multipart)

Attach files directly to POST /v1/predictions/upload. No separate upload step needed.

curl -X POST "https://api.core.today/v1/predictions/upload" \
  -H "X-API-Key: cdt_your_api_key" \
  -F "model=runwayml/gen4-image" \
  -F 'input={"prompt":"your prompt here"}' \
  -F "file:reference_images=@your_file.png"

Image URL

Pass a publicly accessible URL directly. You can also use file_url from the Storage API.

{
  "model": "runwayml/gen4-image",
  "input": {
    "prompt": "your prompt here",
    "reference_images": ["https://example.com/image.jpg"]
  }
}
See the File Upload docs for more upload methods including Presigned URLs.

Common Parameters

Common parameters used when calling POST /v1/predictions.

ParameterTypeRequiredDefaultDescription
modelstringYes-Model identifier
inputobjectYes-Object containing the model-specific parameters from the table above
output_folderstringNo-Folder path for output files (max 256 chars, '..' not allowed)
webhook_urlstringNo-Webhook URL to call on completion
is_publicbooleanNofalseIf true, output files are also available via permanent public URLs

Examples

Two Characters in a Referenced Location

Three references (a park, a woman, a man) combined with @tag mentions

curl -X POST "https://api.core.today/v1/predictions" \
  -H "Content-Type: application/json" \
  -H "X-API-Key: cdt_your_api_key" \
  -d '{
  "model": "runwayml/gen4-image",
  "input": {
    "prompt": "a close up portrait of @woman and @man standing in @park, hands in pockets, looking cool. She is wearing her pink sweater and bangles.",
    "resolution": "1080p",
    "aspect_ratio": "4:3",
    "reference_tags": [
      "park",
      "woman",
      "man"
    ],
    "reference_images": [
      "https://example.com/park.jpg",
      "https://example.com/woman.jpg",
      "https://example.com/man.jpg"
    ]
  }
}'

Tips & Best Practices

1Choose 720p (150 credits) for drafts and iterations, then re-run keepers at 1080p (240 credits)
2Give each reference image a tag and mention it as @tag_name in the prompt โ€” untagged references are harder to control
3Reference images must have an aspect ratio between 0.5 and 2; crop extreme panoramas before uploading
4Fix the seed to iterate on wording while keeping composition stable
5Only use photos of real people as references with the person's explicit consent

Use Cases

Multi-character scenes that keep every person consistent
Product shots placing the same item in new environments
Storyboards and comics reusing characters across panels
Brand campaigns anchored to a consistent visual style
Placing a known character into a specific referenced location