MiniMax H3

Video playground

Hydrating page from local storage...

Generate 5-15 second clips with MiniMax H3, prompt expansion, optional start and end images, and an optional target soundtrack. 2K and 4K upscale a native 768P result.

Requests use the asynchronous fal.ai queue. Pausing polling preserves the existing request for resumption. Model API documentation and pricing.

Choose a workflow

Text to video Image to video

Connection

FAL on Foundry is not available yet. FAL.ai Direct uses Authorization: Key and https://queue.fal.run. Browser deployments use the configured same-origin relay unless direct CORS is enabled.

API provider

API base URL

fal.ai API key

Model endpoint

Advanced

Model controls

Aspect ratio

Duration

Resolution

Seed (optional)

Prompt expansion

Enable safety checker Inline base64 output

Remove C2PA metadata

Only enable metadata removal when your provider agreement allows it. Provenance review retains the original generated bytes.

Your H3 video workspace is ready

Choose a workflow, configure your connection, and add a prompt and any source media. Progress and the finished clip will appear here.

Estimated cost Unavailable[2K]