Vidu Q4 Preview: Release, Specs and What Comes Next
Everything known so far about the first public version of Vidu Q4 — what is available now, what is still missing, and what is not yet announced.
Released October 7, 2026 · Full release date not yet announced
Vidu Q4 Preview at a Glance
Vidu Q4 Preview is the first public release of ShengShu Technology's Vidu Q4 model, launched on October 7, 2026. ShengShu describes it as a preview: the model is open for real projects, but the company has said a full Vidu Q4 release will follow, without giving a date.
"Preview" here means early access to a finished-enough model, not a closed beta. Anyone can generate with it through Vidu's own products, through Vidu's API, and through third-party platforms such as this one. What a preview label does signal is that behavior, limits and prices may still change before the final version ships.
If you want the background on the model itself — what it is and who it is for — start with our Vidu Q4 overview. This page focuses on the release: what the preview can do today, what is still uncertain, and how to plan around it.
- Developer
- ShengShu Technology
- Release
- October 7, 2026 (public preview)
- API model ID
- viduq4-preview
- Modes
- Image to video, reference to video
- Reference images
- Up to 15 (official); up to 12 on this site
- Full release
- Announced, no date given
What You Can Do with Vidu Q4 Preview Today
The preview is not a feature-limited teaser. Its two generation modes are open in full, and both run at every resolution the model supports.
Animate a single image
Image to video turns one picture into a clip and keeps its original aspect ratio. It is the quickest way to see what the preview does with your own material.
Build scenes from references
Reference to video combines reference images and up to 3 voice references into a new scene, with lip-synced speech and multi-shot storytelling in a single generation. Vidu Q4 officially supports up to 15 reference images; this site accepts up to 12 per generation.
Render up to 4K
Output ranges from 540p to 4K, and the 2K and 4K settings use 10-bit color. Clips can run up to 16 seconds, with native audio turned on or off.
Use it through an API
ShengShu exposes the preview on its open platform under the model ID viduq4-preview. That is how independent tools, including this site, offer it in the browser.
Not included in the preview
- Pure text-to-video: every generation needs at least one image.
- Fixed public pricing for the final model: current prices are launch prices.
- A published roadmap or changelog for the full release.
The fastest way to judge it is to run your own test: open the Vidu Q4 video generator and start with a short 540p clip.
Vidu Q4 Preview Specs
These are the limits of the preview as offered on this site, which runs on the fal.ai Vidu Q4 endpoints. They follow the official model specifications except for reference images: Vidu Q4 officially supports up to 15, and this site accepts up to 12. ShengShu may adjust the limits before the full release.
| Spec | Image to Video | Reference to Video |
|---|---|---|
| Input images | 1 image (first frame) | 1–12 reference images (official model limit: 15) |
| Voice references | — | Up to 3 MP3 clips, 3–12 s each |
| Clip length | 3–16 seconds | 3–16 seconds |
| Resolution | 540p, 720p, 1080p, 2K, 4K | 540p, 720p, 1080p, 2K, 4K |
| Color depth | 10-bit at 2K and 4K | 10-bit at 2K and 4K |
| Aspect ratio | Same as the uploaded image | 16:9, 9:16, 1:1, 4:3, 3:4 |
| Native audio | Always on | On or off, with voice cloning from references |
| Prompt | Optional | Required |
For mode-by-mode detail, read the Vidu Q4 image to video and Vidu Q4 reference to video guides; the Vidu Q4 video generator page explains every setting step by step.
Vidu Q4 Preview vs the Full Release
ShengShu has confirmed that a full Vidu Q4 release will follow the preview, but it has not published what will change. Below is what can and cannot be said today.
What we know
- The preview already includes the headline features: multi-image references (officially up to 15; up to 12 on this site), 3 voice references, 16-second clips and 4K output with 10-bit color.
- Both modes are available to everyone through Vidu's products and its API, not only to invited testers.
- Videos generated with the preview are regular outputs you can download and use under the platform's terms.
What is not yet known
- Whether the full release adds text-to-video, which the preview does not support.
- Whether limits such as clip length, resolution or the number of references will change.
- Final pricing, and whether preview launch prices will last.
- Any changes in generation speed or output quality.
What "preview" means for stability
Preview models can change without much notice. In practice that can mean slightly different results for the same prompt and seed over time, longer queues at busy hours, and occasional failed generations. On this site a failed generation never costs credits — they are returned automatically — and finished videos are kept in your account, so a model update does not affect clips you already made.
How to prepare your projects for the full release
- 01Keep a project folder with every reference image, voice clip and final prompt, so any shot can be regenerated later without guessing what went into it.
- 02Write down the mode, resolution, duration and seed for clips you plan to reuse. Small changes to any of them change the result.
- 03Upload the highest-quality source images you have. Clean, sharp references keep characters and products consistent across model versions.
- 04Before committing to 4K renders for a whole series, test one critical shot at the final resolution and length to confirm it holds up.
When Will the Full Vidu Q4 Be Available?
There is no official date. At launch, ShengShu said only that the full model would follow the preview. Any specific date you see quoted elsewhere is a guess unless it comes from Vidu directly.
Should you wait? For most work, no. The preview is the same model family with its main features already enabled, and waiting only gives up the head start of learning a new model early. If you produce a long series that must look identical from episode to episode, keep your reference images and prompts saved, so you can regenerate shots consistently if the model changes.
Oct 7, 2026
Vidu Q4 Preview released
First public version, with image-to-video and reference-to-video.
To be announced
Full Vidu Q4 release
Confirmed by ShengShu; timing and changes not yet published.
When the full model becomes available through the API, we plan to offer it in the Vidu Q4 video generator and will update this page with the confirmed differences.
Vidu Q4 Preview FAQ
What is Vidu Q4 Preview?
It is the first public version of Vidu Q4, the AI video model from ShengShu Technology. It turns one image, or a set of reference images and voice clips, into videos of up to 16 seconds at up to 4K. "Preview" signals that a full release is planned and details may change.
What is the Vidu Q4 Preview release date?
Vidu Q4 Preview was released on October 7, 2026. That date applies to the preview only; ShengShu has not announced when the full Vidu Q4 release will arrive.
Vidu Q4 Preview vs Pro: is there a Vidu Q4 Pro?
No Vidu Q4 Pro has been announced. Vidu has used Pro and Turbo names for earlier generations on its API, so a similar split for Q4 is possible, but nothing is confirmed. Today the model is offered as Vidu Q4 Preview, under the API model ID viduq4-preview.
Is Vidu Q4 Preview free to use?
The model itself is paid per generation, because rendering video uses a lot of computing power. On this site, new accounts get free credits to test it with a short 540p clip, and after that you pay with credits based on resolution and length.
Is Vidu Q4 Preview good enough for real projects?
Yes, as long as you plan for change. ShengShu positions Vidu Q4 for AI series, advertising, social content and cinematic production, and the preview already has the features those need. Save your reference images, prompts and seeds, so you can recreate or extend a project consistently if the model is updated.