Video

Video is our users’ favorite Studio feature. In the app, open Video, pick a live model, describe the clip, and Generate. Studio queues the job, waits for the MP4, then lets you play and download it.

Tap Improve Prompt before you run if you want a sharper take. It has two modes: Pro and Standard. Pro uses Opus 4.8 for the best possible results.

Video models

Studio keeps the latest Venice video models on offer (privacy-first upstream). Drive every one of them from the Video UI: pick a model, fill in what it asks for, and Generate.

There are three kinds of video model. The picker filters them the same way (prompt2video, image2video, reference2video).

Recommended

The picker surfaces a Recommended set at the top. Those picks move with what is hot in the market right now, so do not treat them as fixed. Names and availability change as Venice ships new models too.

Prompt to video

Text in, clip out. You write a prompt (or tap Improve Prompt for a sharper take) and Generate. No image upload. Best when you are inventing a scene from scratch.

Image to video

Start from a still. Upload a source image, describe the motion you want, and Generate. The model treats that image as the starting frame of the clip.

Reference to video

Guide the look with a reference image. Upload a still for style or identity, write a prompt for the action, and Generate. Use this when you want the clip to follow a look without locking that still as the opening frame.

Using an agent or a harness? See Agents video.