fal

name	fal
description	Generate images and videos via Fal.ai API. Use when generating AI images (FLUX, nano-banana-2), AI videos (Veo 3.1), or any Fal.ai model. Supports queue-based async generation with polling, reference image editing, and file download. Also use when the content pipeline needs direct image/video generation bypassing Dropspace rate limits.

Fal.ai Generation

Generate images and videos directly via Fal.ai Queue API. Zero dependencies (Node.js only).

Auth

export FAL_KEY="your-fal-api-key"
export FAL_KEY="your-fal-api-key"

Script

All generation via scripts/fal-generate.js (relative to this skill directory).

Generate Image

node ~/.openclaw/skills/fal/scripts/fal-generate.js image \
  --prompt "founder working at desk, cinematic lighting" \
  --aspect 9:16 \
  --out /tmp/slide.png

Options:

• --prompt (required): Image description
• --aspect: Ratio (default 9:16). Options: 1:1, 16:9, 9:16, 4:3, 3:4, 5:4, 4:5
• --model: Fal model ID (default fal-ai/nano-banana-2). Other: fal-ai/flux/dev, fal-ai/flux/schnell, fal-ai/flux-pro/v1.1-ultra
• --reference-image: URL of reference image (auto-uses /edit endpoint for style guidance)
• --out: Download result to path
• --no-wait: Return request_id immediately without polling
• --timeout: Poll timeout in ms (default 120000)

Generate Video

node ~/.openclaw/skills/fal/scripts/fal-generate.js video \
  --prompt "young founder picks up phone, looks surprised at screen" \
  --aspect 9:16 \
  --duration 8 \
  --out /tmp/clip.mp4

Options:

• --prompt (required): Video description
• --aspect: 9:16 (default) or 16:9
• --duration: 4, 6, or 8 seconds (default 8)
• --model: Fal model ID (default fal-ai/veo3.1)
• --out: Download result to path
• --no-wait: Return request_id immediately
• --timeout: Poll timeout in ms (default 300000)

Note: Duration must be integer (not string). Sora 2 supports 4, 8, or 12 seconds only.

Check Status / Get Result

# Check if a job is done
node ~/.openclaw/skills/fal/scripts/fal-generate.js status \
  --model fal-ai/nano-banana-2 --request-id <id>

# Fetch completed result (+ optional download)
node ~/.openclaw/skills/fal/scripts/fal-generate.js result \
  --model fal-ai/nano-banana-2 --request-id <id> --out /tmp/img.png

Output Format

JSON to stdout. Images:

{"request_id":"...","model":"fal-ai/nano-banana-2","images":[{"url":"https://...","width":768,"height":1344}],"seed":123}

Videos:

{"request_id":"...","model":"fal-ai/veo3.1","video":{"url":"https://...","duration":8}}

Models & Pricing

Model	Type	Cost	Notes
fal-ai/nano-banana-2	Image	~$0.08	Default. Same model Dropspace uses.
fal-ai/nano-banana-2/edit	Image	~$0.08	Reference image guidance (auto-selected with --reference-image)
fal-ai/flux/schnell	Image	~$0.003	Fast, lower quality
fal-ai/flux/dev	Image	~$0.025	Good balance
fal-ai/flux-pro/v1.1-ultra	Image	~$0.05	Up to 2K, best realism
fal-ai/veo3.1	Video	~$0.15/sec	Veo 3.1. 4-8s clips with audio.

Pipeline Integration

To bypass Dropspace's 5/min rate limit, generate images directly then attach to launches:

1. Generate image: fal-generate.js image --prompt "..." --out /tmp/slide1.png
2. Upload to Dropspace launch as pre-made media (no /media/generate call needed)

This avoids checkMediaGenerateLimit entirely since you're uploading finished images, not requesting generation through the Dropspace API.