CLI

The glabs CLI provides terminal-based access to all SDK features — image generation, video generation, project management, Whisk, and an OpenAI-compatible server.

Installation

npm install -g @getvrex/glabs-cli

Verify installation:

glabs --version

Configuration

The CLI reads from ~/.glabs/config.json with environment variable overrides.

Quick Setup

# Extract tokens automatically
glabs auth extract --email [email protected] --password "pass" --save
 
# Or set tokens manually
glabs config set --bearer-token "tok" --session-token "stok" --account-tier pro

Set Config Values

glabs config set --bearer-token "your-bearer-token"
glabs config set --session-token "your-session-token"
glabs config set --account-tier pro
glabs config set --recaptcha-provider chrome
glabs config set --recaptcha-api-key "your-api-key"
glabs config set --output-dir ./output

Multiple values in one command:

glabs config set \
  --bearer-token "your-token" \
  --session-token "your-session-token" \
  --account-tier pro \
  --recaptcha-provider chrome

View Config

glabs config show

Displays all values with sensitive tokens masked.

Environment Variable Overrides

Environment variables take precedence over the config file:

Variable	Config Key
`GLABS_BEARER_TOKEN`	`bearerToken`
`GLABS_SESSION_TOKEN`	`sessionToken`
`GLABS_ACCOUNT_TIER`	`accountTier`
`GLABS_PROJECT_ID`	`projectId`
`GLABS_RECAPTCHA_PROVIDER`	`recaptchaProvider`
`GLABS_RECAPTCHA_API_KEY`	`recaptchaApiKey`
`GLABS_WHISK_COOKIE`	`whiskCookie`
`GLABS_OUTPUT_DIR`	`outputDir`
`GLABS_GOOGLE_EMAIL`	`googleEmail`
`GLABS_GOOGLE_PASSWORD`	`googlePassword`
`GLABS_BROWSERLESS_TOKEN`	`browserlessToken`

Commands

Auth

auth extract

Extract bearer + session tokens via browser automation:

glabs auth extract --email [email protected] --password "pass" --save
glabs auth extract --email [email protected] --password "pass" --browserless-token "tok"

Option	Description
`--email <email>`	Google email (or `GLABS_GOOGLE_EMAIL`)
`--password <password>`	Google password (or `GLABS_GOOGLE_PASSWORD`)
`--browserless-token <tok>`	Browserless.io token (omit for local browser)
`--save`	Save tokens to `~/.glabs/config.json`

Images

images generate

glabs images generate -p "A sunset over mountains" -a 16:9
glabs images generate -p "A cute robot" -n 4 -a 1:1 -o ./my-images
glabs images generate -p "A landscape" -m nanobanana2 --seed 42

Option	Description
`-p, --prompt`	Image prompt (required)
`-a, --aspect-ratio`	`16:9` \| `9:16` \| `1:1` (default: `1:1`)
`-n, --count`	Number of images 1-4 (default: 1)
`-m, --model`	`nanobanana` \| `nanobananapro` \| `nanobanana2`
`--seed`	Random seed
`--reference-media-id`	Reference media ID (repeatable)
`-o, --output-dir`	Output directory (default: `./output`)
`--json`	Output raw JSON

images upload

glabs images upload -f photo.jpg

Option	Description
`-f, --file`	Image file path (required)
`--json`	Output raw JSON

images upsample

glabs images upsample --media-id "abc123" --resolution 4K

Option	Description
`--media-id`	Media ID to upsample (required)
`--resolution`	`2K` \| `4K` (default: `4K`)
`-o, --output-dir`	Output directory
`--json`	Output raw JSON

images credits

glabs images credits

Videos

videos generate (text-to-video)

glabs videos generate -p "A cinematic drone shot" -a 16:9
glabs videos generate -p "A cityscape" --mode fast

Option	Description
`-p, --prompt`	Video prompt (required)
`-a, --aspect-ratio`	`16:9` \| `9:16` \| `1:1` (default: `16:9`)
`--mode`	`quality` \| `fast`
`--account-tier`	Override tier
`--seed`	Random seed
`--json`	Output raw JSON

videos i2v (image-to-video)

glabs videos i2v -p "Camera pans slowly" --start-media-id "abc123"
glabs videos i2v -p "Motion" --start-media-id "abc" --end-media-id "xyz"

Option	Description
`-p, --prompt`	Video prompt (required)
`--start-media-id`	Start frame media ID (required)
`--end-media-id`	End frame media ID (first+last frame mode)
`-a, --aspect-ratio`	`16:9` \| `9:16` \| `1:1`
`--seed`	Random seed

videos r2v (reference-images-to-video)

glabs videos r2v -p "A scene with these characters" --ref-media-id abc --ref-media-id def

Option	Description
`-p, --prompt`	Video prompt (required)
`--ref-media-id`	Reference image media ID (required, repeatable, 1-3)
`-a, --aspect-ratio`	`16:9` \| `9:16` \| `1:1`
`--mode`	`quality` \| `fast`
`--seed`	Random seed

videos extend

glabs videos extend --media-id "abc123" -p "Continue the scene with a sunset"

Option	Description
`--media-id`	Video media ID (required)
`-p, --prompt`	Extension prompt (required)
`-a, --aspect-ratio`	`16:9` \| `9:16` \| `1:1`

videos reshoot

glabs videos reshoot --media-id "abc123" --motion-type RESHOOT_MOTION_TYPE_FORWARD

Option	Description
`--media-id`	Video media ID (required)
`--motion-type`	Camera motion type (required)
`-a, --aspect-ratio`	`16:9` \| `9:16` \| `1:1`

Motion types (all prefixed RESHOOT_MOTION_TYPE_):

Camera: UP, DOWN, LEFT_TO_RIGHT, RIGHT_TO_LEFT, FORWARD, BACKWARD, DOLLY_IN_ZOOM_OUT, DOLLY_OUT_ZOOM_IN_LARGE
Stationary: STATIONARY_UP, STATIONARY_DOWN, STATIONARY_LEFT_LARGE, STATIONARY_RIGHT_LARGE, STATIONARY_DOLLY_IN_ZOOM_OUT, STATIONARY_DOLLY_OUT_ZOOM_IN_LARGE

videos upsample

glabs videos upsample --media-id "abc123" --resolution 4k

Option	Description
`--media-id`	Video media ID (required)
`--resolution`	`1080p` \| `4k` (default: `4k`)

videos status

glabs videos status --operation-name "operations/xyz"
glabs videos status --media-id "abc123"

videos poll

Polls until completion and auto-downloads the video:

glabs videos poll --operation-name "operations/xyz" -o ./videos
glabs videos poll --media-id "abc123"

Option	Description
`--operation-name`	Operation name
`--media-id`	Media ID
`--project-id`	Project ID
`-o, --output-dir`	Output directory (default: `./output`)

Projects

glabs projects list
glabs projects list --page-size 50 --json
glabs projects get <project-id>

Whisk

Generate images via Google Whisk API (Imagen 3.5):

glabs whisk generate -p "A cute robot" -a LANDSCAPE -o ./whisk-output

Option	Description
`-p, --prompt`	Image prompt (required)
`-a, --aspect-ratio`	`SQUARE` \| `PORTRAIT` \| `LANDSCAPE` (also accepts `1:1`, `9:16`, `16:9`)
`--seed`	Random seed
`-o, --output-dir`	Output directory
`--json`	Output raw JSON

Requires whiskCookie set in config: glabs config set --whisk-cookie "cookie"

Serve (OpenAI Server)

Start an OpenAI-compatible HTTP server:

glabs serve
glabs serve --port 3000 --host 127.0.0.1 --api-key "sk-my-key" --max-concurrent 8

Option	Description
`--port`	Listen port (default: `8000`)
`--host`	Bind host (default: `0.0.0.0`)
`--api-key`	Optional API key for auth
`--max-concurrent`	Max concurrent tasks (default: `4`)

Endpoints: POST /v1/chat/completions, GET /v1/models

Typical Workflow

# 1. Setup
glabs auth extract --email [email protected] --password "pass" --save
 
# 2. Generate an image
glabs images generate -p "A mountain landscape" -a 16:9
 
# 3. Upload for video use
glabs images upload -f output/image-1709123456.png
 
# 4. Generate video from image
glabs videos i2v -p "Camera slowly pans right" --start-media-id "abc123"
 
# 5. Poll and download
glabs videos poll --operation-name "operations/xyz" -o ./videos
 
# 6. Upscale to 4K
glabs videos upsample --media-id "def456" --resolution 4k

Output Files

Default output dir: ./output (or --output-dir / config outputDir).

Type	Filename Pattern
Images	`image-{timestamp}.png`
Videos	`video-{timestamp}.mp4`
Whisk	`whisk-{timestamp}.png`
Upsampled	`upsample-{timestamp}.png`

Use --json on any command to output raw JSON to stdout instead.

CLI

On this page