Shengshu: Vidu Q3

Vidu Q3 focuses on reference-driven video creation. It is useful for bringing recognizable characters or visual subjects into a new scene while planning how the accompanying audio contributes to the clip.

Compare models

viduq3

AvailableShengshuVideoAsync

Price: From $0.035
Modalities: Video

About Vidu Q3

Vidu Q3 is Shengshu's third-generation video model in its reference-driven form: you supply recognizable characters or visual subjects and a prompt, and it places them in a new scene. The Q3 generation is the one Vidu documents with simultaneous audio and video output, so sound is part of how the clip is planned. The base Q3 model offers reference-to-video only.

Where it works well

Reference-to-video keeps a character or subject recognizable when moved into a new setting.
The Q3 generation is documented with synchronized audio and video output, so planning sound is part of the workflow.
Clip length is billed by the second, which suits short pieces of varying length.
It is the plain reference-to-video model, simpler than the Mix variant that combines many elements.

When to choose another model

It has no text-to-video or image-to-video operation; use Q3 Pro or Turbo for those.
Every clip needs reference images, so it is a poor fit for quick text-only ideas.
Subject fidelity depends on the quality and clarity of the references you provide.

Getting started

Create API key
Create a key in Console, then use it with every model on the platform.

Send your first request

Copy the example for your language and run it against the endpoint.

Reference to videoTokenLab endpoint

POST/v1/videos/generations

curl -X POST "https://api.tokenlab.sh/v1/videos/generations" \
  -H "Authorization: Bearer sk-xxx" \
  -H "Content-Type: application/json" \
  -d '{
  "operation": "reference-to-video",
  "output_audio": true,
  "model": "viduq3",
  "prompt": "Cinematic sunrise over a calm lake with gentle camera motion.",
  "image_urls": [
    "https://example.com/reference.jpg"
  ]
}'

Parameters

Reference to video

Parameter	Values and limits	Default	Notes
`audio`	true, false	Not provided	Not provided
`duration`	Not provided	Not provided	This field is accepted. The current public contract does not specify a complete value range or list. No additional values are promised. The omitted-field default is not verified.
`image_url`	Not provided	Not provided	The current public contract does not specify an input count limit for this field. Do not infer a count from the example request.
`image_urls`	Inputs: ≥ 1, ≤ 7	Not provided	Image aliases share a total of one to seven reference images. Do not add the limits together.
`output_audio`	true, false	Not provided	Stored route declarations specify different defaults. No uniform omitted-field default is guaranteed; set this field explicitly.
`prompt`	Not provided	Not provided	Not provided
`reference_images`	Inputs: ≥ 1, ≤ 7	Not provided	Image aliases share a total of one to seven reference images. Do not add the limits together.
`resolution`	Not provided	Not provided	This field is accepted. The current public contract does not specify a complete value range or list. No additional values are promised. The omitted-field default is not verified.

Reference-to-video audio defaults to true. Set output_audio or audio to false for silent output.

Pricing

The Verified price applies to Verified, which costs less on most models. The Official price is the model maker's published price and applies to the more reliable Official route. Auto bills the route that completes the request.

Reference to video · 540p

per second

Official price: $0.035
Verified price: —
Discount: —

Reference to video · 720p

per second

Official price: $0.06
Verified price: —
Discount: —

Reference to video · 1080p

per second

Official price: $0.075
Verified price: —
Discount: —

	Official priceper second	Verified priceper second	Discount
Reference to video · 540p	$0.035	—	—
Reference to video · 720p	$0.06	—	—
Reference to video · 1080p	$0.075	—	—

Usage & activity

Success rate is the share of requests that completed. Latency is how long a full response takes; P95 means 95% of requests finished within that time.

Usage & availability

Last 24 hours

Requests
Success rate
P95 latency
Total tokens

Model performance

Last updated Nov 1, 2026, 8:48 AM UTC

Metrics appear once privacy and data volume thresholds are met.

Data is based on aggregate user requests, excluding status checks.

Use cases

Recurring characters
Keep the same animated mascot or presenter across a series of short clips.
Brand subject in new settings
Place a product or logo character into different locations for campaign variants.
Audio-aware shots
Plan clips where dialogue or ambient sound is part of the intended result.
Pre-visualization with cast
Test how a defined character behaves in scenes taken from a script, before investing in a full production.

Prompt examples

The knight from the reference stands on a cliff at dawn and raises the sword, wind in the cape, cinematic wide shot.

The cartoon fox from the reference opens a cafe door and greets a customer, cheerful ambient music.

The robot from the reference repairs a lamp post in the rain, close-up, soft neon reflections.

FAQ

What is Vidu Q3?

It is Shengshu's third-generation Vidu video model. The base model works from reference-to-video: you supply subject images plus a prompt and get back a short clip.

How is Q3 different from Q3 Pro?

Q3 Pro adds text-to-video, image-to-video, and start-to-end frame inputs. The base Q3 is limited to reference-driven generation from your own supplied images and a written prompt.

Does Vidu Q3 generate sound?

Yes. Vidu's model map says Q3 models output audio and video together, so describe the sound you want in the prompt alongside the visuals of the scene.

How is Q3 different from Q2?

Q3 is the newer generation, documented with audio and longer clips. Q2 is a silent series with shorter maximum clips and more input types per tier.

What should my references look like?

Use clear, well-lit images where the subject is easy to see. Avoid clutter and conflicting angles, since the model has to infer the subject's look from them.

How much does Vidu Q3 cost?

On TokenLab, Vidu Q3 costs $0.035-$0.075 per second. The pricing table above shows the full breakdown. Rates depend on the billing unit, specification, and usage. Compare matching conditions in the model's detailed pricing; a single rate does not determine the total cost.

Which endpoint should Vidu Q3 use?

Use https://api.tokenlab.sh/v1/videos/generations for Vidu Q3. The request example below shows the matching code shape.

Which operations does Vidu Q3 support?

Vidu Q3 supports Reference to video. Select an operation above to see its endpoint and request example.