Vidu Q4 Preview: API Pricing, Capabilities and Early Reviews

Vidu Q4 Preview is ShengShu Technology's next-generation flagship preview, not the final Q4 release. Vidu's official update log dates its introduction to October 7, 2026. The API model ID is viduq4-preview. It supports image-to-video and reference-to-video, synchronized audio, up to 15 reference images and three reference audio clips, and output options from 540p to 4K. The current endpoint specifications accept 3–16 seconds; there is no documented Q4 Preview text-to-video or first-and-last-frame endpoint.

The strongest reasons to test it are character performance, reference consistency and cinematic movement—not an assumption that every clip will be production-ready. Its early video-arena result is promising, but audio alignment, extreme action and style consistency still deserve individual acceptance checks.

APIMaster had no public active route for viduq4-preview on October 9, 2026. The Vidu price below is an official API specification price, not an APIMaster selling price. If you need a video route today, compare the available Seedance 2.0 family, Grok Imagine Video and Kling models:

Model ID	Original/reference price	Current APIMaster price	Active routes	Discount	Model card
viduq4-preview	RMB 0.28125/s at 540p; RMB 0.75/s at 1080p, official credit pricing	No public route	0	—	Check availability
seedance-2.0	$0.1512/s, 720p	From $0.14364/s	4	up to ~5% off	Seedance 2.0
seedance-2.0-fast	$0.12096/s, 720p	From $0.06048/s	3	up to ~50% off	Seedance Fast
seedance-2.0-mini	$0.0756/s, 720p	From $0.02268/s	3	up to ~70% off	Seedance Mini
grok-imagine-video-1.5	$0.08/s, marketplace reference	From $0.064/s	2	up to ~20% off	Grok video
kling-v3-omni	$0.084/s, marketplace base reference	From $0.07392/s, base mode	1	up to ~12% off	Kling Omni
kling-v3-motion-control	$0.1286/s, marketplace reference	From $0.10288/s	1	up to ~20% off	Kling Motion Control

Prices and route counts were checked on October 9, 2026, using only routes with status=1; the live marketplace card is authoritative. Discounts compare the same route's price with its listed reference rate, not unlike resolutions or modes. RMB and USD figures are not directly compared. Mode, audio, resolution and reference-video inputs can change the bill.

What is Vidu Q4 Preview, and when was it released?

Vidu Q4 Preview is the first public preview of the next flagship generation. The official update log records it on October 7, 2026, with emphasis on acting, cinematic camera language, complex effects, voice references and more efficient creation. A preview is a version to evaluate now; it does not establish the final Q4 model's release date or feature set.

Use Vidu Q4 Preview as the product name and viduq4-preview in an API request. Do not substitute guessed IDs such as vidu-q4, viduq4 or viduq4-pro.

This article separates official specifications, independent preference testing and a published hands-on review. APIMaster has not run its own Q4 Preview generation benchmark; the observations below are not our original sample tests.

Vidu Q4 Preview capabilities: 4K, audio and reference inputs

The model map and the two Q4 Preview endpoint pages define the current integration boundary:

Capability	Current documented behavior
Image-to-video	One starting image; POST /ent/v2/img2video
Reference-to-video	1–15 images; POST /ent/v2/reference2video
Reference audio	Up to three MP3 clips, each 3–12 seconds, in reference-to-video
Audio output	audio: true enables dialogue and sound effects; false produces silent video
Resolution	540p, 720p, 1080p, 2K and 4K
Frame rate	24fps in the model map
Duration	3–16 seconds in both endpoint specifications and the pricing table
Aspect ratio	Reference-to-video accepts 1:1, 9:16, 16:9, 3:4 and 4:3
Text-only generation	No Q4 Preview text-to-video endpoint documented
First-and-last-frame generation	No dedicated Q4 Preview endpoint documented
Video references	Not exposed as a Q4 Preview request input

The overview map says 1–16 seconds, but the endpoint schemas specify 3–16. Use the endpoint range when building a request. Likewise, do not describe every output as inheriting its source image's ratio: reference-to-video has an explicit aspect_ratio field. Features offered elsewhere in Vidu's product suite are not automatically capabilities of this model.

How much does Vidu Q4 Preview cost?

There are three different numbers a reader may encounter: a web-product introductory offer, the official API credit schedule, and a third-party provider's rate. They are not interchangeable.

Official API credit pricing

Vidu's official API pricing, checked October 9, 2026, lists RMB 0.03125 per credit. Image-to-video and reference-to-video share the following Q4 Preview schedule:

Resolution	Credits per generated second	RMB per second	RMB for a 10-second clip
540p	9	0.28125	2.8125
720p	19	0.59375	5.9375
1080p	24	0.75	7.50
2K	38	1.1875	11.875
4K	78	2.4375	24.375

These RMB amounts are arithmetic conversions of the published credits, not currency-exchange estimates. Vidu states that failed generations, including moderation rejection, do not consume credits. A technically successful but creatively unusable clip is a different outcome: budget for rejected takes, not just failed tasks.

Introductory offers and third-party prices

An introductory web-product offer of RMB 0.09/s appears in launch coverage and the hands-on review linked below. Do not use it as the universal API list price or assume it applies to 4K. Eligibility, resolution, account credits and offer expiry need to be checked at the relevant checkout.

On October 9, the Artificial Analysis I2V leaderboard displayed $5.04/min for Q4 Preview. That is the board's listed provider price, not APIMaster pricing and not an unconditional quote for all resolutions. It also differs from earlier circulating figures such as $7.20/min; use a dated source rather than silently combining them.

For available APIMaster alternatives, the observed route ranges were $0.14364–$0.1512/s for Seedance 2.0, $0.06048–$0.12096/s for Fast, $0.02268–$0.0756/s for Mini, and $0.064–$0.12/s for Grok Imagine Video 1.5. The checked single routes were $0.07392/s for Kling Omni base mode and $0.10288/s for Kling Motion Control. Check the mode-specific card before comparing budgets.

Is Vidu Q4 Preview better than Seedance 2.0?

The Artificial Analysis AA-Video-I2V v1.0 board, checked October 9, 2026, ranked Q4 Preview third in its with-audio view. It showed Elo 1179 ±10, a 95% interval of 1169–1189, and 5,543 samples, explicitly marked provisional. MiniMax H3 Max was 1195, MiniMax H3 1181 and Dreamina Seedance 2.0 720p 1176. Vidu Q3 Pro was 1056, ranked nineteenth.

This is encouraging evidence of improvement over Q3 Pro, but it is not proof of a decisive advantage over Seedance 2.0. Their confidence intervals overlap, and a mixed-resolution preference board does not isolate acting, lip sync, motion reliability or production throughput. A strong average preference result cannot tell you how many retries your particular shot requires.

Use the ranking to choose candidates for a controlled test, not to declare a universal winner. Keep source images, prompts, duration, audio requirements and the evaluation rubric fixed, then compare identity preservation, motion continuity, dialogue alignment and cost per accepted clip.

What do early reviews say about Vidu Q4 Preview?

A published iFanr/APPSO hands-on review tested emotional dialogue, action scenes, moving-camera shots and effects with early access. It found camera direction and subject stability promising, but abrupt emotional changes had unnatural pacing; style control and hallucinations also needed improvement. Its examples were built with deliberate reference preparation, not just one-line prompts. This is useful qualitative evidence, not a measured success rate or a broad community consensus.

The practical lesson is to assign each reference a role: identity, clothing, environment or action target. More images are not automatically better; conflicting references can make the intended result harder to judge.

Public long-form creator evidence is still limited this close to release. Lip sync, audio timing, skin detail, extreme motion and hand-drawn style should stay on your test checklist, rather than becoming categorical claims about every output. Official support for synchronized audio is a capability declaration—not a guarantee that dialogue always aligns. Do not base a production purchase on reposted praise or complaints without the original clip and request settings.

How to call the Vidu Q4 Preview API

Use a Vidu-issued API key for this official-provider example. It is not an APIMaster endpoint, and an APIMaster key is not interchangeable with it.

The official image-to-video documentation specifies Authorization: Token, a one-image input and asynchronous task creation:

curl --request POST 'https://api.vidu.com/ent/v2/img2video' \
  --header "Authorization: Token ${VIDU_API_KEY}" \
  --header 'Content-Type: application/json' \
  --data '{
    "model": "viduq4-preview",
    "images": ["https://example.com/your-reference-image.jpg"],
    "prompt": "The character notices a distant sound, turns slowly, then smiles. Keep identity and clothing consistent.",
    "is_rec": false,
    "duration": 5,
    "resolution": "720p",
    "audio": true
  }'

Replace the example image URL with a publicly accessible asset you have permission to use. Save the returned task_id, then query the task with the same Vidu key:

curl --request GET "https://api.vidu.com/ent/v2/tasks/${TASK_ID}/creations" \
  --header "Authorization: Token ${VIDU_API_KEY}"

Wait for state: "success" before using the output URL. Handle queueing, processing and failure separately, and impose a polling timeout. This request is based on the documented schema; it is not presented as an APIMaster-run generation test.

For multi-image or voice conditioning, use the reference-to-video endpoint, with images, optional sounds, and aspect_ratio. Verify its current authentication and limits before integration; do not copy unrelated Vidu features into the payload.

Verdict: where Seedance, Grok and Kling fit

Vidu Q4 Preview is worth testing for reference-led character scenes, expressive camera work and effects-heavy short clips. Its documented controls and provisional preference score make it an interesting candidate, but neither a discounted trial nor a leaderboard position establishes final-release stability.

For a video workflow you can access through APIMaster today, shortlist models by the task instead of substituting one provider's name into another provider's request:

• Seedance 2.0: a practical comparison baseline; include Fast and Mini when screening cost-sensitive drafts. Read the Seedance API guide for the available family and exact IDs.
• Grok Imagine Video 1.5: another active provider to compare with the same prompt and source assets; use its own request schema and acceptance criteria.
• Kling V3 Omni: compare its documented generation modes rather than treating the base starting price as the audio or 4K price.
• Kling Motion Control: a separate choice when a motion-reference workflow is required. It is not the same model or billing contract as Omni, and these available V3 IDs are not a claim that Kling 4.0 is already routed here.

Test short clips first. Track generation time, accepted-take rate and billed seconds, including the route's rules for reference inputs. A cheaper generated second is only useful if enough of those seconds survive editing.

Start with an available model: register on APIMaster, create an API key, set the base URL to https://apimaster.ai/v1, and select seedance-2.0, seedance-2.0-fast, seedance-2.0-mini, grok-imagine-video-1.5, kling-v3-omni or kling-v3-motion-control. Use the video API documentation for the correct model-specific payload. The checked Mini route offers up to ~70% off its listed reference rate; the live card controls current pricing. For Q4 Preview itself, use Vidu's official platform until an active APIMaster route is confirmed.

FAQ

What is the official Vidu Q4 Preview model ID?

viduq4-preview. The display name is Vidu Q4 Preview; vidu-q4-api is this article's URL slug, not an API model identifier.

Is this the final Vidu Q4 release?

No. It is the preview version recorded in Vidu's official update log on October 7, 2026. Do not assume the final version will have the same identifier or limits.

Does Vidu Q4 Preview support text-to-video?

The current Q4 Preview documentation exposes image-to-video and reference-to-video, not a text-only or dedicated first-and-last-frame endpoint. Start with an image.

Can it generate 4K video with sound?

The documented resolution choices include 4K, and audio: true enables audio-video output. Validate the selected mode's output, audio alignment and invoice before a batch.

Is RMB 0.09 per second the official API price?

No. That introductory web-product price is a different offer. The checked official API credit table starts at 9 credits/s for 540p, equivalent to RMB 0.28125/s at RMB 0.03125 per credit.

Can I call Vidu Q4 Preview through APIMaster now?

The October 9, 2026 check found zero public active routes for viduq4-preview. Check its model card for updates; use the active Seedance, Grok or Kling IDs above if you need an APIMaster video route today.

Sources and verification

• Vidu official update log: October 7 release entry and product positioning.
• Vidu model map, image-to-video, and reference-to-video: current model ID and request boundaries.
• Vidu API pricing: credit conversion, resolution schedule and failed-task billing.
• Artificial Analysis I2V leaderboard: with-audio, provisional ranking, uncertainty and displayed provider pricing.
• iFanr/APPSO hands-on review: early-access qualitative observations, not APIMaster's own tests.
• APIMaster live marketplace endpoints for the model IDs in the opening table, checked October 9, 2026. Only status=1 routes were counted; availability and pricing can change.