runway
Generate and transform video and images with Runway models via inference.sh CLI. Models: Gen-4.5, Gen-4 Turbo, Aleph 2.0, Act-Two, Gen-4 Image, Gen-4 Image Turbo. Capabilities: text-to-video, image-to-video, 2-10s duration, video-to-video transformation with keyframe guidance, character performance transfer from a reference video, text-to-image with tagged reference images. Use for: cinematic clips, animating stills, restyling footage, reframing aspect ratio, character animation, consistent characters across images. Triggers: runway, runwayml, runway gen 4, gen-4, gen 4.5, gen-4.5, gen-4 turbo
Try your first task
A source-grounded starter task has not been verified for this skill yet.
SKILL.md
name: runwaydescription: "Generate and transform video and images with Runway models via inference.sh CLI. Models: Gen-4.5, Gen-4 Turbo, Aleph 2.0, Act-Two, Gen-4 Image, Gen-4 Image Turbo. Capabilities: text-to-video, image-to-video, 2-10s duration, video-to-video transformation with keyframe guidance, character performance transfer from a reference video, text-to-image with tagged reference images. Use for: cinematic clips, animating stills, restyling footage, reframing aspect ratio, character animation, consistent characters across images. Triggers: runway, runwayml, runway gen 4, gen-4, gen 4.5, gen-4.5, gen-4 turbo, runway aleph, aleph 2, act-two, act two, runway act, runway image, gen-4 image, runway video, performance transfer"allowed-tools: Bash(belt *)Install the belt CLI skill:
npx skills add belt-sh/cli
Runway Video and Image Generation
Generate, transform and animate video and images with Runway models via inference.sh CLI.
Quick Start
Requires inference.sh CLI (
belt). Install instructions
belt login
belt app run runway/gen-4-5 --input '{"prompt": "a paper boat drifting down a rain-soaked city gutter, macro shot"}'
Models
| Model | App ID | Best For |
|---|---|---|
| Gen-4.5 | runway/gen-4-5 | Highest quality text-to-video and image-to-video, 2-10s |
| Gen-4 Turbo | runway/gen-4-turbo | Fast image-to-video (image required) |
| Aleph 2.0 | runway/aleph-2 | Video-to-video transformation with text and keyframe guidance |
| Act-Two | runway/act-two | Animate a character image or video with a reference performance |
| Gen-4 Image | runway/gen-4-image | Text-to-image with optional tagged reference images |
| Gen-4 Image Turbo | runway/gen-4-image-turbo | Fast image generation guided by a required reference image |
Runway apps use pixel ratios (1280:720) instead of 16:9. Aleph 2.0 is the exception: its target_aspect_ratio uses 16:9 style values.
Search Runway Apps
belt app search "runway"
Examples
Text-to-Video (Gen-4.5)
Text-to-video supports 1280:720 or 720:1280 only.
belt app run runway/gen-4-5 --input '{
"prompt": "a lone hiker crossing a glacier at sunset, wide drone shot, slow push in",
"ratio": "1280:720",
"duration": 10
}'
Image-to-Video (Gen-4.5)
Image-to-video adds 1104:832, 832:1104, 960:960 and 1584:672.
belt app run runway/gen-4-5 --input '{
"prompt": "the cat turns its head toward the window, curtains moving in the breeze",
"image": "https://cat-photo.jpg",
"ratio": "960:960",
"duration": 5
}'
Fast Image-to-Video (Gen-4 Turbo)
belt app run runway/gen-4-turbo --input '{
"prompt": "slow orbit around the product, soft studio light",
"image": "https://product.jpg",
"ratio": "720:1280",
"duration": 5
}'
Transform a Video (Aleph 2.0)
belt app run runway/aleph-2 --input '{
"video": "https://street-clip.mp4",
"prompt": "turn the scene into a snowy winter evening with warm street lights"
}'
Reframe and Guide with Keyframes (Aleph 2.0)
Up to 5 keyframes, each with a reference image at a timestamp. range sets how many seconds around the timestamp the guidance applies.
belt app run runway/aleph-2 --input '{
"video": "https://interview.mp4",
"prompt": "restyle as a hand-painted watercolor animation",
"target_aspect_ratio": "9:16",
"keyframes": [
{"seconds": 0, "image": "https://watercolor-style-frame.jpg"},
{"seconds": 4, "image": "https://watercolor-style-frame-2.jpg", "range": 1}
]
}'
Performance Transfer (Act-Two)
The character mimics the performance in reference_video (3-30s, max 16MB).
belt app run runway/act-two --input '{
"character": "https://cartoon-character.png",
"character_type": "image",
"reference_video": "https://me-talking.mp4",
"body_control": true,
"expression_intensity": 4,
"ratio": "1280:720"
}'
Use "character_type": "video" to drive a character that is already in a video (max 16MB).
Text-to-Image (Gen-4 Image)
belt app run runway/gen-4-image --input '{
"prompt": "editorial photo of a vintage espresso machine on a marble counter",
"ratio": "1920:1080"
}'
Tagged References (Gen-4 Image)
Tag reference images and use @tagName in the prompt to keep characters or objects consistent.
belt app run runway/gen-4-image --input '{
"prompt": "@maya sitting in @cafe reading a book, morning light",
"reference_images": [
{"image": "https://maya-portrait.jpg", "tag": "maya"},
{"image": "https://cafe-interior.jpg", "tag": "cafe"}
],
"ratio": "1360:768"
}'
Fast Image Edit (Gen-4 Image Turbo)
belt app run runway/gen-4-image-turbo --input '{
"prompt": "the character standing on @rooftop at night, city lights behind",
"image": "https://hero-character.jpg",
"additional_references": [{"image": "https://rooftop.jpg", "tag": "rooftop"}],
"ratio": "1024:1024"
}'
Workflow: Consistent Character to Video
# 1. Create a keyframe with a tagged character reference
belt app run runway/gen-4-image --input '{
"prompt": "@maya on a train platform in the rain",
"reference_images": [{"image": "https://maya-portrait.jpg", "tag": "maya"}],
"ratio": "1280:720"
}'
# 2. Animate the keyframe
belt app run runway/gen-4-5 --input '{
"prompt": "the train arrives, her coat moves in the wind, she looks up",
"image": "<image-url-from-step-1>",
"ratio": "1280:720",
"duration": 8
}'
Parameters (Gen-4.5)
| Parameter | Type | Default | Description |
|---|---|---|---|
prompt | string | required | Text description, max 1000 chars |
image | file | - | Start image for image-to-video |
ratio | enum | 1280:720 | 1280:720, 720:1280 (text-to-video); also 1104:832, 832:1104, 960:960, 1584:672 with an image |
duration | integer | 5 | 2-10 seconds |
seed | integer | - | Reproducible results |
Gen-4 Turbo takes the same fields, with image required.
Parameters (Aleph 2.0)
| Parameter | Type | Default | Description |
|---|---|---|---|
video | file | required | Input video, max 16MB |
prompt | string | - | Transformation guidance |
target_aspect_ratio | enum | - | 16:9, 4:3, 3:2, 1:1, 2:3, 3:4, 9:16, 21:9 |
keyframes | object[] | - | Up to 5 {seconds, image, range} entries |
seed | integer | - | Reproducible results |
Parameters (Act-Two)
| Parameter | Type | Default | Description |
|---|---|---|---|
character | file | required | Image (max 5MB) or video (max 16MB) |
character_type | enum | image | image, video |
reference_video | file | required | Performance video, 3-30s, max 16MB |
ratio | enum | 1280:720 | 1280:720, 720:1280, 960:960, 1104:832, 832:1104, 1584:672 |
body_control | boolean | false | Enable non-facial movement and gestures |
expression_intensity | integer | - | 1-5, higher is more expressive |
seed | integer | - | Reproducible results |
Parameters (Gen-4 Image / Turbo)
| Parameter | Type | Default | Description |
|---|---|---|---|
prompt | string | required | Use @tagName to reference tagged images |
image | file | required (Turbo only) | Reference image that Turbo transforms |
reference_images | object[] | - | Gen-4 Image: {image, tag} entries |
additional_references | object[] | - | Turbo: extra {image, tag} entries |
ratio | enum | 1280:720 | 1280:720, 720:1280, 720:720, 960:720, 720:960, 1680:720, 1920:1080, 1080:1920, 1024:1024, 1080:1080, 1168:880, 1360:768, 1440:1080, 1080:1440, 1808:768, 2112:912 |
seed | integer | - | Reproducible results |
Pricing
| App | Price |
|---|---|
runway/gen-4-5 | $0.12/sec |
runway/gen-4-turbo | $0.05/sec |
runway/aleph-2 | $0.28/sec |
runway/act-two | $0.05/sec |
runway/gen-4-image | $0.05/image (720p), $0.08/image (1080p) |
runway/gen-4-image-turbo | $0.02/image |
Estimate before running: belt app estimate runway/gen-4-5 --input input.json.
Sample Workflow
# 1. Generate sample input to see all options
belt app sample runway/gen-4-5 --save input.json
# 2. Edit the prompt
# 3. Run
belt app run runway/gen-4-5 --input input.json
Related Skills
# Full platform skill (all apps)
npx skills add inference-sh/skills@infsh-cli
# All video generation models
npx skills add inference-sh/skills@ai-video-generation
# Image-to-video workflows
npx skills add inference-sh/skills@image-to-video
# Video prompting
npx skills add inference-sh/skills@video-prompting-guide
# Other video models
npx skills add inference-sh/skills@kling
npx skills add inference-sh/skills@pixverse
npx skills add inference-sh/skills@seedance
npx skills add inference-sh/skills@google-veo
Browse all video apps: belt app list --category video
Documentation
- Running Apps - How to run apps via CLI
- Streaming Results - Real-time progress updates
- Content Pipeline Example - Building media workflows