Fal#

<!-- TODO: expand -->

Fal is an inference platform optimised for low-latency image and video generation. JSON2Video integrates with Fal-hosted models for fast image generation (especially Flux variants) and short video clips.

When to use Fal#

• You want the lowest latency for image generation at render time.
• You're using a Flux-family model and want Fal's optimised inference pipeline.
• You need a fast turnaround on video generation models hosted on Fal.

How it appears in JSON2Video#

Fal-backed models are listed in the AI models catalog. Reference them on image or video elements:

{
  "type": "image",
  "prompt": "Minimalist tech product on a soft pastel background",
  "model": "flux-pro",
  "duration": 4
}

Check the live AI models catalog for the current list of Fal-backed models.

Bring your own Fal API key#

To pay Fal directly and use any model available in your Fal account:

1. Get an API key from fal.ai/dashboard/keys.
2. Create a Connection at json2video.com/dashboard/connections of type Fal and paste the key.
3. Reference the Connection by ID in the element:

{
  "type": "image",
  "prompt": "...",
  "model": "flux-pro",
  "connection": "my-fal"
}

Cost#

JSON2Video-billed usage is documented in the credit consumption table. When using a Connection, Fal bills you directly for inference per their pricing page.

See also#

• AI models catalog
• Image element reference
• Flux models
• Dashboard — Connections