Vidu Q3 Pro vs Turbo: How to Run a Fair First Test

Kadraris offers two Vidu Q3 variants: Vidu Q3 Pro and Vidu Q3 Turbo. The words "Pro" and "Turbo" do not tell you which one suits your shot. The only way to know is a fair test: one goal, one source, and settings that both forms accept. This page is the method, not a verdict. It reports no results.

What it costs to run

The sample request uses Vidu Q3 Pro, currently from $0.142 per second; Turbo is from $0.081 per second. Those are starting prices, so check your settings on the pricing page or in the calculator.

• No subscription, and the balance has no expiry.
• Failed generations are free.
• A successful clip is billed even if it loses the test.

Set the rules first

• One prompt and a pass condition. For example: "the object moves smoothly and does not change shape".
• Shared settings. Pick a duration and resolution that both models offer.
• One reference file. Use the same image for both. Some aspect-ratio behavior follows the image input, so a different file would skew the test.
• Real cost, not headline cost. Note each estimate with your matching settings before generating.

A starting request

The body for POST /v1/generate. If you add a reference image, give a URL the API can reach. For the second run, change only model.

{
  "model": "vidu-q3-pro",
  "input": {
    "prompt": "A yellow rubber duck drifting across a calm pond, locked camera, pale morning light, no text"
  }
}

Score without knowing which is which

1. Hide the labels. Rename the files so you do not know which model made which.
2. Score separately: faithfulness to the reference, quality of motion, and how usable the clip is in an edit.
3. Keep cost apart from quality. Write the price next to each score, not into it.
4. Repeat only if the result is unclear. One good take does not promise a whole series.

Before you share conclusions, make sure:

• Prompt and source were identical for both runs.
• Quality and cost were scored separately.
• Any speed claim comes from a timed run, not from the variant's name.

Questions

Can I compare them using different reference images?

You can, but the test stops being fair. Aspect ratio and other behavior can follow the input image, so differences may come from the files rather than the models.

Does this cover Vidu Q4?

No. Only the Q3 variants in the catalog are covered, and nothing is claimed about models that are not listed.

Where are all Vidu Q3 Pro inputs listed?

On the Vidu Q3 Pro page and in the generate docs. Studio and estimated_cost show the price before you send.