Vidu Q4 Preview breaks into the top three for image-to-video generation: pricing stays the same

Better quality at the same price. Artificial Analysis builds its rankings from blind pairwise human evaluations: participants choose between videos without knowing the model names. Vidu charges the same 24 credits per second of 1080p video with audio as it does for Q3 Pro. At the standard rate, that is $7.20 per minute via the API.

Q4 Preview generates videos lasting 3–16 seconds with audio and resolutions up to 4K. Reference-to-Video accepts up to 15 images and up to 3 audio references. Each audio file must be an MP3 and last 3–12 seconds.

Getting started in the browser. On the Vidu Q4 page, the Generate Now button opens the generation interface. Image-to-Video requires one image and a description of the motion. Reference-to-Video accepts images and audio references.

To integrate it into an app, the documentation provides a ready-to-use cURL request: Image-to-Video uses POST /ent/v2/img2video, and Reference-to-Video uses POST /ent/v2/reference2video. The request specifies model: viduq4-preview and a key in the Authorization: Token <API_KEY> header. Image-to-Video also requires images and prompt.

By default, the API generates 5 seconds at 720p with audio. Retrieve the result via GET /ent/v2/tasks/{id}/creations. The video link is valid for 24 hours, so the app must download the file within that time.

Vidu released Preview ahead of the full Q4 version and added the model to its official changelog on Oct 7.