Vidu Q4 API pricing: what 540p to 4K video costs per second

Vidu Q4 is ShengShu Technology's video model, released as Vidu Q4 Preview on October 7, 2026, and live on UnificAlly from $0.0297/sec at 540p up to $0.2674/sec at 4K. Every video comes with native audio, runs 1 to 16 seconds, and starts from at least one image: there is no text-to-video mode. This post is the Vidu pricing breakdown for developers: the full per-second rate card, what a real request costs, where the cheap seat is, and what Q4 adds over Q3. For parameters and a playground, see the Vidu Q4 API page.

TL;DR: Vidu Q4 costs $0.0297–$0.2674/sec of output on UnificAlly, billed by the second, with the same rate for image-to-video and reference-to-video. A 6-second 1080p presenter ad costs $0.5346; a 16-second 4K video costs $4.2784. Our pick: draft at 540p and render the final at 1080p or 4K, and skip 720p, which sits close to the 1080p rate for a visibly softer picture. Pay per use, no subscription, and the $0.50 signup balance covers about 16 seconds of 540p output to test the prompt syntax before you spend anything.

How much does the Vidu Q4 API cost?

Vidu Q4 is billed per second of finished video, and only the resolution changes the rate. The number of reference images (up to 15), audio references (up to 3), the aspect ratio and the video_format encoding do not change the price. Rates below refresh from the live price list on every deploy.

Failed tasks are not billed; you pay for completed videos only. Image-to-video needs at least 3 seconds, so the cheapest possible request is a 1-second reference-to-video at 540p. The cheapest useful request is a 3-second 540p image-to-video at $0.0891.

Vidu cost per second: worked examples

The table is the rate; these are the invoices. Each row is one request at the duration developers actually ask for.

Two things fall out of these numbers. First, the resolution ladder is steep at the top: compare the 1080p and 4K rows, and 4K belongs on the final deliverable, never on the draft. Second, 720p ($0.0713/sec) is the tier to skip. It costs most of what 1080p ($0.0891/sec) costs, while 540p ($0.0297/sec) is a fraction of either and keeps the same motion and audio. Draft at 540p until the prompt and the reference order are right, then rerun the identical request at 1080p, 2K or 4K. Every resolution works in both modes, so nothing else in the request changes.

Vidu Q4 vs Q3: what the new price buys

Vidu Q3 already produced up to 16 seconds of video with native audio and spoken dialogue, and Q4 keeps all of that. ShengShu's Q4 Preview announcement lists what is new: 2K and 4K output with 10-bit color, up to 15 reference images for characters, wardrobe, props and environments in one setup, up to 3 audio references so a character keeps the same voice across videos, and multi-shot scenes where the camera cuts between angles inside one generation.

On the API, the practical change is that the 540p to 1080p rates buy the whole Q4 feature set: a request with one image at 1080p costs the same whether or not it uses the new references or multi-shot prompting. The 2K and 4K rows are the only new line items, and they only matter when the output has to be delivered at that size. ShengShu has not published pricing for the final Q4 release; this post follows the preview rates the API serves today.

Is Vidu Q4 worth it? What our tests showed

Our house tests on the Vidu Q4 page are the evidence behind the pick above. A 6-second 1080p reference-to-video with a presenter photo as [Image1] and an apple as [Image2] kept her face, hair and jacket, kept the apple's exact striped skin, and delivered the spoken line in the audio track. That is the product-ad workflow, and at $0.5346 per take it is the request we would build on. A 5-second image-to-video from a portrait kept the photo's framing and spoke the line, with small, pose-bound motion. A one-reference drone move over an alpine ridge rendered a smooth camera path but ignored the golden-hour light the prompt asked for, so treat lighting instructions as a wish, not a setting.

Where Q4 disappoints is scope: there is no text-to-video, no start-and-end frame control, and no video references or video editing. If the brief starts from words alone, Vidu Q4 is the wrong model whatever its price.

How Vidu Q4 pricing compares with other video models

Per second of 1080p output on UnificAlly today:

As of this writing Vidu Q4 carries the lowest 1080p rate in that set, and it is the only one of the four that goes to 4K. For a text-only prompt, HappyHorse 1.1 API is the closest priced model with text-to-video; for videos longer than 16 seconds, Seedance 2.5 pricing covers the 30-second option.

How to call Vidu Q4 at the price you planned

The cheapest way to call Vidu Q4 is the same way as the most expensive one: one API key, POST /v1/tasks, and resolution set to the tier you priced above. This is the 6-second 1080p presenter ad from the examples table, billed at $0.5346 when it completes.

curl -X POST https://api.unifically.com/v1/tasks \
  -H "Content-Type: application/json" \
  -H "Authorization: Bearer YOUR_API_KEY" \
  -d '{
    "model": "vidu/q4",
    "input": {
      "prompt": "[Image1] stands in a bright kitchen, holds up the bottle from [Image2], smiles and says: \"Cold-pressed, nothing added.\"",
      "reference_image_urls": [
        "https://example.com/host.jpg",
        "https://example.com/bottle.jpg"
      ],
      "resolution": "1080p",
      "duration": 6,
      "aspect_ratio": "16:9"
    }
  }'

Swap reference_image_urls for a single image_url to run image-to-video, which keeps the input image's shape and needs duration of 3 or more. Change resolution to 540p for the draft and to 4K for the final; nothing else in the body moves. The response returns a task_id; poll GET /v1/tasks/<task_id> or set callback_url. Full field reference: docs.unifically.com/models/video/vidu/q4.

Vidu Q4 FAQ

How much does the Vidu Q4 API cost?

Vidu Q4 costs $0.0297/sec at 540p, $0.0713/sec at 720p, $0.0891/sec at 1080p, $0.1783/sec at 2K and $0.2674/sec at 4K on UnificAlly, billed per second of output. Image-to-video and reference-to-video share the same rate, and failed tasks are not billed.

What does a 6-second 1080p Vidu Q4 video cost?

$0.5346. A 16-second 4K video costs $4.2784, and a 3-second 540p draft costs $0.0891.

Do reference images or audio references change the price?

No. Up to 15 reference images and 3 audio references are included in the per-second rate, and so are the aspect ratio and the video_format encoding. Only the resolution and the duration change what a request costs.

What is the cheapest way to use the Vidu API?

Draft at 540p, which is the lowest per-second rate, and rerun the same request at 1080p, 2K or 4K only for the final version. New UnificAlly accounts get a $0.50 signup balance, enough for about 16 seconds of 540p output, so you can test the [Image1] and [Audio1] prompt syntax before paying anything.

Do 2K and 4K cost extra on Vidu Q4?

Yes, they are the two highest tiers: $0.1783/sec per second at 2K and $0.2674/sec per second at 4K, against $0.0891/sec at 1080p. The other Q4 additions, 15 reference images, 3 audio references and multi-shot scenes, carry no surcharge at any resolution.

What will change this post

We will update the rate table when the final Vidu Q4 release replaces the preview, when ShengShu publishes separate pricing for a Q4 Turbo or Pro variant, when Vidu Q4 lands on the Arena video board with a rank worth citing, or when a rate change moves it off the lowest 1080p spot in the comparison table. Prices on this page refresh from the live price list on every deploy, so the tokens stay current between those updates.

Related reading on Vidu Q4

• Vidu Q4 API: playground, parameters, house test videos.
• Seedance 2.5 API pricing: the 30-second option when 16 seconds is not enough.
• Kling 4.0 API: source-video editing and the std/pro ladder.