AI models catalog#

JSON2Video integrates with several AI providers to generate assets at render time. Models are addressed by name in three element types:

• image — image element with a prompt and a model.
• video — video element with a prompt and a model.
• voice — voice element with a text and a model (or implicit default).

Models marked featured are the recommended defaults. Models with credits per unit set to 0 do not consume credits when invoked with JSON2Video's managed key; usage may still be subject to per-provider rate limits.

When a connection is supplied on the element, the request runs against the customer's own provider account and no JSON2Video credits are deducted. When no connection is supplied, JSON2Video's managed key is used and credits are deducted according to the table below.

How to use a model#

{
    "type": "image",
    "model": "flux-schnell",
    "prompt": "A futuristic city at sunset",
    "duration": 5
}

{
    "type": "voice",
    "model": "elevenlabs-flash-v2-5",
    "voice": "JBFqnCBsd6RMkjVDRZzb",
    "text": "Welcome to the future of automated video.",
    "connection": "your-elevenlabs-connection-id"
}

Notes#

• Aliases (e.g. text-to-image-basic, nano-banana, elevenlabs) redirect to a concrete model. Use them when you want JSON2Video to pick a reasonable default in your stead.
• Provider-side rate limits apply on top of any JSON2Video quota. Failed AI generation surfaces as a render error; see Errors.
• The credit cost shown here is what JSON2Video deducts when using its managed API keys. If you set a connection on the element, the cost is borne by your provider account instead.
• The catalog above is fetched live from https://cdn.json2video.com/data/ai/models.json and cached for 15 minutes. If the upstream is unreachable, the last cached version is shown.