Vidu Q4 Preview: release, price, vs Q3

ShengShu released Vidu Q4 Preview on 7 October 2026, ahead of the full Q4. It makes 3 to 16 second clips with native audio from one image or up to 15 reference images, at 540p to 4K. Vidu's list price runs from $0.045 a second at 540p to $0.39 at 4K. Versely runs it from 11 credits.

Key facts

Item	Detail
Released	7 October 2026, as a preview ahead of the full Q4 model
Modes	Image to video (one start frame) and reference to video. No text to video or start/end frame in Vidu's docs
References	Up to 15 images and 3 mp3 voice clips (3 to 12 s each), in Vidu's API
Output	3 to 16 seconds at 24 fps, 540p to 4K, with native dialogue and sound effects. Vidu says 2K and 4K are 10-bit
API price, listed 16 October 2026	9 to 78 Vidu credits a second by quality, $0.045 to $0.39 at list
Launch discount	30% off image to video and reference to video, runs to 30 November 2026
On Versely	Two rows from 11 credits. Up to 12 reference images, no voice clips

Vidu's model map lists 1 to 16 seconds; its API reference and pricing table say 3 to 16, the range Versely uses. Vidu says final pricing, resolutions, features and terms may vary by plan and region.

Sources, all read on 16 October 2026: Vidu's Q4 page, image-to-video and reference-to-video API references, model map, pricing and changelog; the PR Newswire release of 7 October and the same day's GlobeNewswire release, read on The Manila Times; Artificial Analysis' image-to-video leaderboard; and fal's listing. The performance claims are Vidu's, not Versely's own tests.

Checking the headline claims

Up to 15 reference images and 3 voice clips. True in Vidu's API. The reference endpoint takes 1 to 15 images (png, jpeg or webp) and 0 to 3 mp3 clips of 3 to 12 seconds each. Hosts can cap lower: fal documents 12 images, and Versely's reference row takes up to 12.

Up to 16 seconds. True. Duration is a whole number from 3 to 16, defaulting to 5, and the model map lists 24 fps.

2K and 4K with 10-bit colour. Both resolutions are real options. The 10-bit claim appears only in the releases: PR Newswire's covers 2K and 4K, GlobeNewswire's only 4K. Vidu's API reference has no bit-depth or codec field and its Q4 page never mentions bit depth, so it stays Vidu's claim.

Third on Artificial Analysis. Reported, but not something we can confirm. AlphaSignal and OrcaRouter both wrote on 8 October 2026 that Q4 Preview entered AA-Video-I2V v1.0 in third place, with Vidu Q3 Pro 19th. On 16 October 2026 Artificial Analysis' page shows a v1.1 board with audio (23 models) and a silent board (82). Neither lists Q4 Preview, and Vidu Q3 Pro is 16th and 13th. A notice there says a v2.1 board with "a more comprehensive model selection" is coming. Neither Vidu release cites a benchmark.

Vidu Q4 Preview price: what Vidu lists

The two launch releases disagree on price. PR Newswire's release says launch pricing "starts at $0.014/sec" and promises "up to five times as much output for the same budget", with no baseline named. GlobeNewswire's release gives $0.045 a second at 540p up to $0.39 at 4K, with image to video and reference to video 30% off through 30 November 2026. Vidu's pricing page agrees with that scale in credits, and fal's listing applies the discount.

Quality	Vidu credits a second	List $ a second	30% off $ a second
540p	9	0.045	0.0315
720p	19	0.095	0.0665
1080p	24	0.12	0.084
2K	38	0.19	0.133
4K	78	0.39	0.273

Credits are from Vidu's pricing page. The GlobeNewswire release states the 540p and 4K dollar rates; the other three come from fal's listing and equal Vidu's credits at $0.005 each (our arithmetic). Even the discounted 540p rate is more than twice $0.014, and we found no rate card or plan that produces that figure.

Vidu Q4 Preview vs Q3: what changed

• More references. Vidu's Q3 reference endpoints take 1 to 7 images. Q4 Preview takes 1 to 15.
• A higher ceiling. Q3 stops at 1080p. Q4 Preview adds 2K and 4K. Q3 Pro also runs to 16 seconds at 24 fps.
• Voice references. The input is not new: up to three mp3 clips also appear in Vidu's Q3 reference docs. What Vidu claims as new is steadier voices and acting.
• Fewer modes. Q3 also offers text to video and start/end frames. Q4 Preview lists image to video and reference to video only.
• Price. At 540p, 720p and 1080p, Vidu lists Q4 Preview at 9, 19 and 24 credits a second, against 9, 20 and 24 for Q3 Pro. 2K and 4K are the new tiers.
• Quality claims. The releases and changelog promise more nuanced expressions, fluid action, in-scene effects and cuts that follow direction. The only outside voice is one creator's testimonial in the GlobeNewswire release.

Who it is for, and who should wait

Good fit:

• Creators who need one character or product to hold across a clip. The reference row takes up to 12 images on Versely.
• Product, social and ad clips that need dialogue and sound effects without a separate audio pass.

Wait, or use another model, if:

• You need text to video or start/end frames. Versely's Vidu Q3 Video covers text to video from 28 credits.
• You need voice references through Versely, or all 15 images. Versely's reference row has no voice input and stops at 12 images.
• You need a fixed spec or price. It is a preview, and Vidu says terms may change.
• You choose by leaderboard. We found no Q4 Preview row to cite.

How to use it on Versely

Versely runs two Vidu Q4 rows, both from 11 credits: Vidu Q4 for image to video and Vidu Q4 Reference for reference to video. There is no text to video, start/end frame or edit row, and one model page covers both. Clips run 3 to 16 whole seconds, at 540p, 720p, 1080p, 2K or 4K. Image to video has no aspect-ratio setting, and Vidu says output keeps the source image's ratio. The reference row offers 16:9, 9:16, 4:3, 3:4 and 1:1. Native audio comes with image to video and is optional on the reference row.

Quality	5 s	10 s	16 s
540p	18	36	58
720p	38	76	122
1080p	48	96	154
2K	76	152	244
4K	156	312	500

Credits per job on 16 October 2026, the same on both rows. A 3-second clip at 540p is the 11-credit minimum, and 16 seconds at 4K is the 500-credit maximum. Audio adds no credits. Billing is per second, rounded up per job. The model page lists whole-credit rates (4 for SD, 8 for HD, 32 for 4K a second); the table has the exact totals.

1. Start from the AI video generator on the web, or in the Versely iPhone app.
2. For image to video, attach your starting image and pick Vidu Q4. For references, pick Vidu Q4 Reference and add up to 12 images.
3. Write the motion, the camera move and the sound. In the reference row, address each image by position, as [@reference_image_1], [@reference_image_2] and so on.
4. Pick the length and quality. Test a prompt at 540p first, the cheapest tier, then spend on 2K or 4K.

The sample on the model page ends its prompt with a sound line: "Sound: rhythmic chugging, a long echoing steam whistle, rushing water." Our guides cover writing motion for a still, preparing a reference set and prompting sound on Vidu Q3, Q4's predecessor.

FAQ

When was Vidu Q4 Preview released?

ShengShu released it on 7 October 2026. Vidu's API changelog, the PR Newswire release and the GlobeNewswire release all carry that date, and Vidu calls it a preview ahead of the full Q4.

How much does Vidu Q4 Preview cost?

Vidu lists 9 to 78 credits a second, which is $0.045 at 540p up to $0.39 at 4K at list, as read on 16 October 2026. A 30% launch discount runs to 30 November 2026. On Versely a job runs from 11 to 500 credits.

Does Vidu Q4 Preview do text to video?

No. Vidu lists image to video and reference to video only. On Versely, Vidu Q3 Video covers text to video.

How many reference images does Vidu Q4 take?

Up to 15 images and 3 voice clips through Vidu's own API. Versely's reference row takes up to 12 images and no voice clips.

What is the Vidu Q4 Preview API model ID?

viduq4-preview, on POST https://api.vidu.com/ent/v2/img2video for image to video and POST https://api.vidu.com/ent/v2/reference2video for reference to video. Result URLs last 24 hours.