vidu-q4Vidu Q4 is the newest video model in ShengShu's Vidu family, released as a Preview (upstream model name viduq4-preview). Vidu describes it as its new flagship: it generates sound together with the picture, including dialogue and sound effects, and it can cut between shots on its own. On APIMODELS it is one model id, vidu-q4, on the unified video endpoint, and requests go to Vidu's official API. Two modes are chosen from what you send. Image-to-video takes one first frame in first_frame_url (image and image_url are accepted too); the clip keeps that image's aspect ratio, the prompt is optional, and is_rec turns Vidu's prompt enhancement on or off (on by default). Reference-to-video takes 1 to 15 images in reference_image_urls, such as characters, products, outfits or scenes, plus up to 3 MP3 clips of 3 to 12 seconds each in reference_audio_urls to guide voices; a prompt is required, you refer to the references by number, and aspect_ratio can be 16:9 (default), 9:16, 1:1, 4:3 or 3:4. Both modes take duration as any whole number of seconds from 3 to 16 (default 5), resolution 540p, 720p (default), 1080p, 2K or 4K, generate_audio (default true), and an optional seed. There is no text-only mode, no last frame and no reference video; those requests return a clear 400 instead of being silently dropped. Output has no watermark. Billing is per second of the length you request, and audio does not change the price: 540p $0.042/s, 720p $0.0887/s, 1080p $0.112/s, 2K $0.1773/s and 4K $0.3639/s, so a 5-second 720p clip is $0.4435 and a 16-second 4K clip is $5.8224. Only successful generations are charged; failed ones are refunded.
PNG / JPEG / WebP. The video keeps this image's aspect ratio.
Dialogue, sound effects and ambience are generated with the picture · Same price with or without audio
Vidu rewrites the prompt for richer detail (default)
$0.4435 for this clip (720p, $0.0887/s × 5s) · only successful generations are billed
Generated video will appear here
Provide URLs and click Generate
Dialogue, sound effects and ambience are generated together with the video. Send generate_audio: false for a silent file; the price is the same either way
Reference-to-video keeps characters, products, outfits and scenes consistent across new actions and camera angles. Name each reference by number in the prompt
In reference mode, add MP3 clips of 3-12 seconds in reference_audio_urls to guide how characters sound
Send first_frame_url and an optional prompt; the clip keeps the image's aspect ratio, and is_rec switches Vidu's prompt enhancement on or off
Five resolution tiers and any whole length from 3 to 16 seconds. Reference-to-video adds 16:9, 9:16, 1:1, 4:3 and 3:4
540p $0.042/s, 720p $0.0887/s, 1080p $0.112/s, 2K $0.1773/s, 4K $0.3639/s for the seconds you request. Failed generations are refunded
Vidu Q4 is the newest video model in ShengShu's Vidu family, released as a Preview under the upstream name viduq4-preview. Vidu describes Vidu Q4 as its new flagship: it generates the picture and the sound in one pass, dialogue and sound effects included, and it can cut between shots on its own. Vidu Q4 Preview goes up to 4K, runs 3 to 16 seconds per clip, and accepts up to 15 reference images and 3 reference audio clips to keep characters and voices consistent.
On APIMODELS the Vidu Q4 API is one model id, vidu-q4, on the same POST /v1/video/generations endpoint as every other video model, and each request goes to Vidu's official API. One API key covers Vidu Q4 together with the rest of the catalog, billing is per second from $0.042, and only successful Vidu Q4 generations are charged. You can try Vidu Q4 in the playground on this page, see the exact price of a clip before you generate it, and then copy the same parameters into your own code.
The Vidu Q4 API has two modes, chosen from what you send. Image-to-video animates one first frame. Reference-to-video builds a new scene from up to 15 reference images and up to 3 reference audio clips. Both modes share duration, resolution, native audio and seed, so switching a Vidu Q4 workflow from one mode to the other changes the inputs, not the rest of the request.
Vidu positions Vidu Q4 for character-driven video: faces that change expression, bodies that react, and scenes that move from a quiet conversation to fast action. Because Vidu Q4 generates speech together with the picture, a character can talk on screen without a separate voice-over pass. Describe the performance in the prompt, who speaks, what they say and how they react, and Vidu Q4 plays it out in the same clip.
Send 1 to 15 images in reference_image_urls to control how characters, clothing, products and locations look while Vidu Q4 invents new actions and camera angles. Combine a character portrait, a product shot and an environment photo in one request, then name them in the prompt by position: the first image is reference image 1 (Vidu's own examples write it as 参考图1). A prompt is required in this mode, and aspect_ratio can be 16:9, 9:16, 1:1, 4:3 or 3:4.
With generate_audio on, which is the default, Vidu Q4 returns a clip with dialogue, sound effects and ambience synced to the picture; turn it off for a silent file at the same price. In reference-to-video you can add up to 3 MP3 clips of 3 to 12 seconds each in reference_audio_urls to guide how characters sound, and refer to them as reference audio 1, 2 and 3. That keeps a recurring character's voice recognizable from one Vidu Q4 clip to the next.
Vidu lists automatic shot switching among Vidu Q4's headline features, so one generation can move between angles instead of holding a single shot. Ask for the camera you want in plain words, a slow push-in, a tracking shot behind a running character, a cut to a close-up when someone speaks, and keep the subjects anchored with reference images so they stay the same across cuts. Explosions, smoke, fireworks or dramatic light are prompt material too; judge the result on a 540p draft before paying for a 4K render.
resolution takes 540p, 720p (default), 1080p, 2K or 4K in both modes, and duration takes any whole number of seconds from 3 to 16. Use 540p at $0.042 per second for quick tests, then render the take you keep at 1080p, 2K or 4K; a 5-second 4K clip from Vidu Q4 is $1.8195. Image-to-video keeps the aspect ratio of your first frame, so crop the image to the shape you want before you send it.
Step 1, get your Vidu Q4 API key: sign up at apimodels.app, open the console and create an API key. The same key works for Vidu Q4 and every other model, with one USD balance; top-ups start at $10.
Step 2, test Vidu Q4 in the playground: on this page pick Image to Video or Reference to Video, upload your images (and MP3 voice clips in reference mode), write the prompt, choose duration and resolution, and read the price shown before you click Generate.
Step 3, integrate the Vidu Q4 API: POST https://api.apimodels.app/v1/video/generations with Authorization: Bearer YOUR_API_KEY and model vidu-q4. Send first_frame_url for image-to-video, or reference_image_urls plus a prompt for reference-to-video; the docs page for Vidu Q4 lists every field with cURL, Python and Node examples.
Step 4, retrieve your Vidu Q4 videos: the call returns a taskId. Poll GET /v1/video/generations?task_id= every few seconds until the state is completed, or pass callback_url and we POST the result to you. Result files are kept for 30 days, so download or re-host what you need.
The Vidu Q4 API is billed per second of the length you request, by resolution. Audio on or off is the same price, reference images and reference audio cost nothing extra, and a failed Vidu Q4 generation is refunded.
Eligible new accounts get $0.10 of trial credit, which is less than the cheapest Vidu Q4 clip (540p, 3 seconds, $0.126), so plan on a small top-up for your first test.
| Vidu Q4 resolution | Per second | 5 seconds | 10 seconds | 16 seconds |
|---|---|---|---|---|
| 540p | $0.042 | $0.21 | $0.42 | $0.672 |
| 720p (default) | $0.0887 | $0.4435 | $0.887 | $1.4192 |
| 1080p | $0.112 | $0.56 | $1.12 | $1.792 |
| 2K | $0.1773 | $0.8865 | $1.773 | $2.8368 |
| 4K | $0.3639 | $1.8195 | $3.639 | $5.8224 |
We ran Vidu Q4 through our own API on 2026-10-22, one request per mode at 540p and 3 seconds. Image-to-video took 102 seconds from submission to the finished file; reference-to-video with 2 reference images and 1 MP3 voice reference took 173 seconds. Both Vidu Q4 clips came back as 24 fps H.264 MP4 files with an AAC audio track, which confirms native audio in both modes, with no watermark and a length of 3.04 seconds for a 3-second request.
These are single runs at the lowest tier, not a benchmark: 1080p, 2K, 4K and longer clips were not part of the test and will likely take longer. Plan for minutes rather than seconds, poll every few seconds or pass callback_url, and use FLUX 3 Max if a clip has to be back within seconds.
| Vidu Q4 mode (540p · 3 s) | Inputs | Submit to finished video | Output file |
|---|---|---|---|
| Image-to-video | 1 first frame + prompt | 102 s | 962×536 (follows the first frame), 24 fps, H.264 + AAC, 3.04 s |
| Reference-to-video | 2 reference images + 1 MP3 voice clip + prompt | 173 s | 960×528 (16:9), 24 fps, H.264 + AAC, 3.04 s |
Vidu Q4 Preview has no text-only mode: every Vidu Q4 request needs a first frame or at least one reference image. It has no last frame for start-to-end transitions, no keyframes, no reference video and no editing of an existing clip, and a single clip stops at 16 seconds. Requests that ask for any of these get a clear 400 at create time, so nothing is billed for them.
Pick another model when you hit those limits: FLUX 3 Max for text-to-video, a last frame or up to 10 keyframes; Seedance 2.5 for clips up to 30 seconds, reference videos, editing or extension; Kling V3 Omni for editing a clip while keeping its original sound. Because Vidu Q4 is a Preview, Vidu may still change how it behaves, so pin your own test prompts and re-check them after updates.
One key, many video models: Vidu Q4 sits next to Seedance, Kling, FLUX 3 Max, Wan, MiniMax H3 and the Gemini Omni models on the same endpoint and the same balance, so comparing Vidu Q4 with another model is a change of one field, not a new vendor account.
Prices you can check before you spend: every Vidu Q4 price on this page is the per-second rate we bill, the playground shows the cost of the clip before you generate it, and GET /v1/records/{task_id} returns what a finished Vidu Q4 job actually cost.
Pay as you go, refunds on failure: there is no subscription and no minimum monthly spend, a failed Vidu Q4 generation is refunded automatically, and inputs the model cannot take are rejected with a 400 before anything is charged.
Built for automation: asynchronous tasks with polling or callback_url, the same response shape across video models, and result files stored on our CDN for 30 days, so batch jobs and SaaS pipelines can run Vidu Q4 without managing model infrastructure. The API is reachable from mainland China as well as worldwide.
Vidu Q4 suits short, reference-driven clips where the same characters, products or voices have to come back again and again. Six common ways teams put the Vidu Q4 API to work:
Let users upload a photo and a motion prompt and get a clip back. Image-to-video covers single photos, reference-to-video covers apps where users bring several characters or products, and the five resolution tiers let you offer a cheap preview at 540p and a paid export at 2K or 4K from the same Vidu Q4 integration.
Build a film shot by shot in clips of up to 16 seconds. Reuse the same reference images and voice clips in every request so the cast looks and sounds the same from scene to scene, and let Vidu Q4 generate the dialogue and ambience with each shot.
Turn product photos and brand visuals into promotional clips: put the product, the model and the setting in reference images, describe the camera move, and render vertical 9:16 for social ads or 16:9 for a landing page, all from one Vidu Q4 call per variant.
Animate portraits and illustrations into characters that move and speak. Give Vidu Q4 the character as a reference image and a voice clip as reference audio, write the line in the prompt, and get a talking clip for an app, a lesson or an interactive story. Use only images and voices you have the rights to.
Generate TikTok, Reels and Shorts variations from the same assets by changing the prompt, the aspect ratio or the seed. Because the Vidu Q4 API is asynchronous with callbacks, a content pipeline can queue many jobs and collect the files as they finish.
Add Vidu Q4 video generation to a creative tool, a CMS or a media product: accept the user's images and prompt, submit the task, store the returned file. Per-second pricing makes it straightforward to pass the cost on to your own customers.
All prices below are the per-second rates on apimodels.app today, with audio. Vidu Q4 is the only one of these with 4K output and has the lowest per-second price at 720p with audio; it stops at 16 seconds and needs an image to start from. Seedance 2.5 is the pick for long clips, reference videos and editing, Kling V3 Omni for editing a clip while keeping its sound, and FLUX 3 Max for text-to-video and keyframes.
| Vidu Q4 | Seedance 2.5 | Kling V3 / V3 Omni | FLUX 3 Max | |
|---|---|---|---|---|
| Max resolution | 4K | 720p | Standard / pro tiers | 1080p |
| Clip length | 3-16 s | 4-30 s | 3-15 s | 5-20 s |
| Text-to-video | No | Yes | Yes | Yes |
| Image inputs | 1 first frame, or up to 15 reference images | Up to 30 reference images | First frame (V3); image and subject input (V3 Omni) | First / last frame, or up to 10 keyframes |
| Audio references | Up to 3 MP3 clips | Up to 10 | Voice IDs via voice_list (up to 2) | No |
| Video input / editing | No | Up to 10 videos, edit and extend | V3 Omni: reference video and editing | No |
| Price per second with audio | 540p $0.042 · 720p $0.0887 · 1080p $0.112 · 2K $0.1773 · 4K $0.3639 | 480p $0.120 · 720p $0.270 | V3 $0.18 (standard) · $0.24 (pro); V3 Omni $0.117 · $0.147 | 720p $0.221 · 1080p $0.377 |
Vidu Q4 is a Video Generation API provided by Vidu. Vidu Q4 is the newest video model in ShengShu's Vidu family, released as a Preview (upstream model name viduq4-preview). Vidu describes it as its new flagship: it generates sound together with the picture, including dialogue and sound effects, and it can cut between shots on its own. On APIMODELS it is one model id, vidu-q4, on the unified video endpoint, and requests go to Vidu's official API. Two modes are chosen from what you send. Image-to-video takes one first frame in first_frame_url (image and image_url are accepted too); the clip keeps that image's aspect ratio, the prompt is optional, and is_rec turns Vidu's prompt enhancement on or off (on by default). Reference-to-video takes 1 to 15 images in reference_image_urls, such as characters, products, outfits or scenes, plus up to 3 MP3 clips of 3 to 12 seconds each in reference_audio_urls to guide voices; a prompt is required, you refer to the references by number, and aspect_ratio can be 16:9 (default), 9:16, 1:1, 4:3 or 3:4. Both modes take duration as any whole number of seconds from 3 to 16 (default 5), resolution 540p, 720p (default), 1080p, 2K or 4K, generate_audio (default true), and an optional seed. There is no text-only mode, no last frame and no reference video; those requests return a clear 400 instead of being silently dropped. Output has no watermark. Billing is per second of the length you request, and audio does not change the price: 540p $0.042/s, 720p $0.0887/s, 1080p $0.112/s, 2K $0.1773/s and 4K $0.3639/s, so a 5-second 720p clip is $0.4435 and a 16-second 4K clip is $5.8224. Only successful generations are charged; failed ones are refunded. Through APIMODELS platform, you can access this model via a unified API with transparent pay-as-you-go pricing. Current pricing: 540p · per second: $0.042, 540p · 5s: $0.21, 540p · 10s: $0.42, 720p · per second: $0.0887, 720p · 5s: $0.4435, 720p · 10s: $0.887, 1080p · per second: $0.112, 1080p · 5s: $0.56, 1080p · 10s: $1.12, 2K · per second: $0.1773, 2K · 5s: $0.8865, 2K · 10s: $1.773, 4K · per second: $0.3639, 4K · 5s: $1.8195, 4K · 10s: $3.639.
Related model: FLUX 3 Max — when you need text-to-video or up to 10 keyframes, and clips up to 20 seconds — from $0.221/s
Related model: Seedance 2.5 — for clips up to 30 seconds, reference videos, or editing and extending an existing video — 480p from $0.120/s, 720p $0.270/s
Related model: Kling V3 Omni — when you edit an existing clip and want to keep its original sound — from $0.088/s, $0.117/s with audio
Quickly generate brand promotion videos for ad campaigns and social media marketing.
Create compelling short-form video content for platforms like TikTok, Instagram, and YouTube.
Generate product feature demonstrations and tutorials to improve user conversion.
Produce course explanations, knowledge explainers, and training videos at an affordable price.
Vidu Q4 is available through APIMODELS at: 540p · per second: $0.042, 540p · 5s: $0.21, 540p · 10s: $0.42, 720p · per second: $0.0887, 720p · 5s: $0.4435, 720p · 10s: $0.887, 1080p · per second: $0.112, 1080p · 5s: $0.56, 1080p · 10s: $1.12, 2K · per second: $0.1773, 2K · 5s: $0.8865, 2K · 10s: $1.773, 4K · per second: $0.3639, 4K · 5s: $1.8195, 4K · 10s: $3.639. Billing is pay-as-you-go — you only pay for what you generate.
Sign up at APIMODELS, get your API key, and call our unified API endpoint. We provide detailed API documentation with code examples in cURL, Python, and Node.js.
APIMODELS offers the same Vidu Q4 model through our aggregation platform. We provide a unified API interface so you do not need separate accounts for each provider - one API key to access all models.
Vidu Q4 is the newest video model in ShengShu's Vidu family, currently released as a Preview (upstream model name viduq4-preview). It generates the video and its sound in one pass, including dialogue and sound effects, and according to Vidu it can cut between shots on its own. On apimodels.app it is the model id vidu-q4 on POST /v1/video/generations, with two modes: image-to-video from one first frame, and reference-to-video from up to 15 reference images plus up to 3 reference audio clips, 3 to 16 seconds at 540p up to 4K.
Yes. Vidu Q4 Preview is the model version Vidu has opened on its official API, under the name viduq4-preview, and the Vidu Q4 API on apimodels.app is how you call it, with the model id vidu-q4. Preview is Vidu's own label: the model is new and Vidu may still change how it behaves. If Vidu publishes a newer Q4 version, this page will say which version vidu-q4 runs.
Not entirely. Eligible new accounts get $0.10 of trial credit, but the shortest and cheapest Vidu Q4 clip, 3 seconds at 540p, costs $0.126, so a first test needs a top-up; top-ups start at $10. The playground on this page shows the exact price of the clip before you click Generate, and only successful generations are charged, so a failed attempt costs nothing.
Yes. resolution accepts 540p, 720p (default), 1080p, 2K and 4K in both modes. 4K is $0.3639 per second, so a 5-second 4K clip is $1.8195 and a 16-second one $5.8224; 1080p is $0.112/s and 2K $0.1773/s. A practical workflow is to try prompts and references at 540p ($0.042/s) or 720p ($0.0887/s), then render the version you keep at 2K or 4K.
Any whole number of seconds from 3 to 16, set with duration (default 5). You pay for the seconds you request: a 16-second clip is $0.672 at 540p, $1.4192 at 720p, $1.792 at 1080p, $2.8368 at 2K and $5.8224 at 4K. If one shot has to run longer, Seedance 2.5 goes up to 30 seconds and FLUX 3 Max up to 20 seconds.
Up to 15, in reference_image_urls, as PNG, JPEG or WebP. Use them for the characters, products, outfits and locations that must stay recognizable; the prompt refers to them by position, the first image being reference image 1 (Vidu's own examples write it in Chinese as 参考图1). Image-to-video takes exactly one image, the first frame, and a request cannot mix a first frame with reference images.
Yes, by default. With generate_audio true (the default) Vidu Q4 creates dialogue, sound effects and ambience in sync with the picture; send false for a silent clip, at the same price. In reference-to-video you can also pass up to 3 MP3 clips of 3 to 12 seconds each in reference_audio_urls to guide the voices, and name them in the prompt as reference audio 1, 2 and 3. Reference audio is not available in image-to-video.
Per second of the length you request: 540p $0.042/s, 720p $0.0887/s, 1080p $0.112/s, 2K $0.1773/s and 4K $0.3639/s, so a 5-second clip is $0.21, $0.4435, $0.56, $0.8865 or $1.8195 at those tiers. Reference images and reference audio cost nothing extra, audio on or off is the same price, and only successful generations are charged; failed ones are refunded.
POST https://api.apimodels.app/v1/video/generations with your API key as a Bearer token and model vidu-q4. For image-to-video add first_frame_url, plus an optional prompt and is_rec; for reference-to-video add reference_image_urls and a prompt, plus optional reference_audio_urls and aspect_ratio. duration, resolution, generate_audio and seed work in both modes. The call returns a taskId: poll GET /v1/video/generations?task_id= until the state is completed, or pass callback_url. Result files are kept for 30 days.
Measured via the API on 2026-10-22, one run each at 540p and 3 seconds: image-to-video took 102 seconds from submission to the finished video, and reference-to-video with 2 reference images and 1 MP3 voice reference took 173 seconds. Both returned a 24 fps H.264 MP4 with an AAC audio track and no watermark, 3.04 seconds long. Higher resolutions and longer clips were not part of this test and will likely take longer, so poll every few seconds or use callback_url.
It has no text-only mode, no last frame for start-to-end clips, no reference video or video editing, no keyframes, and clips stop at 16 seconds; such requests get a clear 400 instead of a silently degraded video. For text-to-video or keyframes use FLUX 3 Max, for clips up to 30 seconds with reference videos, editing or extension use Seedance 2.5, and for editing an existing clip while keeping its sound use Kling V3 Omni.
On APIMODELS, Vidu Q4 runs alongside 60+ models on one API key and one balance, so choosing is about fit, not lock-in. It supports Image to Video, Reference to Video, Up to 15 Reference Images, Up to 3 Voice References, Native Audio, 540p to 4K, 3-16s, From $0.042/s, and you can weigh it on price and capability against other Video Generation models, then switch by changing a single model-name string — no new account or integration. Browse every Video Generation option with live pricing at apimodels.app/models.
Vidu Q4 supports: Image to Video, Reference to Video, Up to 15 Reference Images, Up to 3 Voice References, Native Audio, 540p to 4K, 3-16s, From $0.042/s. See the APIMODELS docs for full parameters and call examples.
Yes. APIMODELS exposes Vidu Q4 through a single unified API and one key — no separate provider accounts, and no need to handle each provider's regional network access yourself.
We support Stripe (Visa, Mastercard, and other international cards) and Alipay. Credits are available instantly after payment.