AI Video Generator
Alibaba's first Wan with sound built in: speech, effects and music come out with the picture, from a low-cost 480p tier.
Output
Four generations of Alibaba's Wan video model run on this site, and a newer one is not always dearer. A five second 720p clip costs 126 credits on Wan 2.6 or 2.7, 150 on Wan 2.5 and 180 on Wan 3.0, so the right pick depends on the length, the size and whether you start from a picture.
September 2025, with a low-cost 480p tier
Five or ten second clips at 480p, 720p or 1080p, from text or from a still. At 480p a five second clip costs 75 credits, the lowest price for a five second Wan clip written from text. From 720p up, Wan 2.6 and 2.7 cost less and do more.
December 2025, with the low-cost Flash tier
Text and image to video in 5, 10 or 15 second clips at 720p or 1080p. Its Flash tier animates a still at 720p for 75 credits with sound or 38 without, the lowest price for a five second Wan clip from a picture.
Open the Wan 2.6 pageApril 2026, stills and clips from one model
Text and image to video at any whole second from 2 to 15, at 720p or 1080p, for the same price as Wan 2.6. The same page also makes stills: text to image and image to image at 15 credits each.
Open the Wan 2.7 pageThe newest, up to 30 seconds
Text, image and reference to video, where the reference mode takes images and a motion clip. Any length from 2 to 30 seconds, with a switch for sound. A five second clip is 90 credits at 480p; at 720p and 1080p it costs more than Wan 2.7.
Open the Wan 3.0 pageLow on a wooden stage, a tap dancer's shoes run through a fast rhythm. The clicks in the soundtrack keep time with the steps on screen, because the sound is made in the same pass as the picture, not laid over it afterwards.
A tap dancer in black tap shoes and grey trousers performs a fast rhythm on a wooden stage under a single spotlight, the camera low at floor level on her feet. Every tap lands on the beat. Sound: crisp tap shoe clicks in rhythm, a light echo in the empty theatre, no music.
A photo of a bike mechanic goes in as the first frame with a single line in the prompt. He wipes his hands, grins and tells the customer the chain was bone dry, his mouth moving with the words and his face the one from the photo.
Start frameHe wipes his hands on the rag, looks at the customer just off camera and says with a grin: "Your chain was bone dry. Rides like new now." Then he gives a small nod. The camera holds a steady medium close-up. Sound: his voice, a ratchet clicking on the bench, quiet street noise through the open door, no music.
A square 480p clip of an old wooden record player. The tone arm swings down onto the spinning record, a click and a crackle follow, and a soft jazz piano comes in once the needle is in the groove, all from Wan 2.5's lowest tier.
Square close-up of a vintage wooden record player in warm lamplight. The tone arm lowers by itself, the needle touches the spinning black record and a soft jazz piano begins to play. Slow push in on the needle in the groove. Sound: the click of the arm, a crackle of vinyl, then a warm jazz piano melody.
Rough out a scene at 480p, hear the dialogue and effects, and pay for a larger size only once the idea works.
Give a portrait one line to say and get a short clip of the person saying it.
Kettles, footsteps, a ticking clock: short scenes built around one sound that lands on the right frame.
How the model works, based on public documentation and sourced evidence.
Wan 2.5 is the preview generation of Alibaba's Wan video model family, announced in September 2025. Alibaba describes it as a natively multimodal architecture: instead of a video model with audio bolted on afterwards, a single unified framework is trained jointly on text, image, video, and audio data. Coverage of the launch also notes the team used reinforcement learning from human feedback to improve instruction following and keep sound and picture tightly aligned.
In practical terms, Wan 2.5 generates clips of 5 or 10 seconds at 480p, 720p or 1080p from either a text prompt or a starting image. Because audio is generated in the same pass as the video, dialogue, ambient sound and background music land on the frames they belong to. The model can also follow an uploaded soundtrack and time the picture to it.
The headline change over earlier Wan releases is that native synchronized audio and 1080p output are built in, where previous versions produced silent video that creators had to score and lip sync in post. Alibaba positions this as the fix for the out-of-sync problem common in AI video, with lip movement, sound effects, and music generated to match on-screen motion.
One caveat for anyone coming from the open-source side: Wan 2.1 and 2.2 shipped open weights under the Apache 2.0 license, but Wan 2.5 launched as an API-only preview rather than an open-weights release. As of late 2025 no model weights had been released, and a community request for them on the official Wan GitHub repository remained unanswered.
On Vidney, Wan 2.5 runs text to video and image to video in 5 or 10 second clips at 480p, 720p or 1080p, and every clip comes back with sound; there is no switch to turn it off. Text to video takes 16:9, 9:16 and 1:1 at every size, and 4:3 or 3:4 at 720p and 1080p only, because the 480p tier has no size in those shapes. Image to video keeps the shape of the picture you give it. A five second 480p clip is the cheapest Wan text to video on the site; from 720p up, Wan 2.6 and 2.7 cost less. Uploading your own soundtrack and Wan 2.5's image models are not offered here.
Sources: x.com · wan.video · github.com
Wan 2.5 is the September 2025 release of Alibaba's Wan video model and the first in the family to make sound and picture in the same pass. On Vidney it turns a prompt or a photo into a 5 or 10 second clip at 480p, 720p or 1080p, and every clip comes back with sound: lines spoken on the right lips, effects on the frames that cause them, music when you ask for it. The reason to pick it today is the 480p tier, the lowest price for a five second Wan clip written from text, which makes it the one for drafts, rough ideas and volume. From 720p up, Wan 2.6 and Wan 2.7 cost less and take longer clips, so the final cut usually moves to them.
Create AI video with Wan 2.5 on Vidney in three steps.
Specs side by side with similar models, all runnable on Vidney.
| Wan 2.5 | Vidu Q4 Preview | Grok Imagine Video 1.5 | Grok Imagine Video 1.5 Lite | |
|---|---|---|---|---|
| Task modes | Text to Video · Image to Video | Image to Video · Reference to Video | Text to Video · Image to Video · Reference to Video | Text to Video · Image to Video |
| Duration | 5, 10 seconds | 3–16 seconds | 1–15 seconds | 1–15 seconds |
| Resolution | 480p · 720p · 1080p | 540p · 720p · 1080p · 2K · 4K | 480p · 720p · 1080p | 480p · 720p · 1080p (upscaled from 720p) |
| Audio | Included, always on | Supported | Included, always on | Included, always on |
| Cost | 75 credits per generation | 159 credits per generation | 18 credits per generation | 18 credits per generation |
Recurring themes from public reviews, comparisons and creator write-ups. Both the good and the bad.
Creators consistently praise the native audio as the standout feature: hands-on reviews found it generates dialogue, fitting sound effects, and cinematic background music in a single pass, with lip sync that in good runs matches the audio almost perfectly, making it a favorite for talking-head and music-driven videos.
Testers rate its handling of complex, action-heavy prompts highly; in one multi-model comparison against Kling 2.5 and Veo 3, Wan 2.5 followed a demanding scene brief with the highest accuracy and excelled at dynamic motion and atmosphere.
Reviewers also point to value: at its lowest resolution it is one of the cheaper ways to get a clip that already has sound, and many called it the more practical choice over Veo 3 for everyday work.
A common complaint is inconsistency: generated voices can sound robotic or glitchy, and creators report needing multiple retries before lip sync and character stability come out right, with reviewers noting Veo 3 still feels more reliable for dialogue.
The open-source community is frustrated that Wan 2.5 broke the family's pattern of open releases; unlike Wan 2.1 and 2.2 it is API-only, and requests for the weights on the official GitHub repository went unanswered.
More AI video models you can run on Vidney.
Alibaba's Wan 2.6 with its Flash tier: the lowest price per second in the Wan family for putting a still in motion.
Up to 30 seconds in one pass, sound included: write the scene, animate a start frame, or cast it from reference images.
Alibaba's Wan 2.7 makes the frame and then moves it: stills, edits and clips with sound from one model family.
Generate AI video with Kling 2.5 Turbo: text to video and image to video, all from one Vidney workspace.
Generate AI video with Seedance 2.0: text to video and image to video, all from one Vidney workspace.
Generate AI video with Kling 2.6: text to video and image to video, all from one Vidney workspace.
Wan 2.5 is an AI video model you can run on Vidney. It supports text to video and image to video from a single workspace.
Wan 2.5 costs 75 credits per generation. The exact credit cost is shown on the generate button before you run it. That is about $2.25 per run at the Starter plan rate ($0.03 per credit).
Wan 2.5 supports the following aspect ratios: 1:1, 4:3, 3:4, 16:9, 9:16.
Wan 2.5 generates clips of 5, 10 seconds.
Yes. Wan 2.5 generates video with sound in the same pass: dialogue, effects and room tone. There is no switch to turn it off.
Not with sign-up credits alone. New accounts get 10 to 30 free credits, and Wan 2.5 costs 75 credits per generation, so running it needs a plan. You can spend the free credits on lower-cost models first; the exact cost is always shown on the generate button before you spend a credit.
Yes. The videos you generate with Wan 2.5 on Vidney are yours to use in commercial projects: ads, social posts, client work, and product content. You keep the output; Vidney only handles generation and storage.
Only on the free plan. Free-plan downloads carry a small Vidney watermark; on any paid plan, Wan 2.5 videos download clean, ready to publish or hand to a client. Upgrading also removes the watermark from videos you already made.
The closest Wan 2.5 alternatives on Vidney are Vidu Q4 Preview, Grok Imagine Video 1.5, and Grok Imagine Video 1.5 Lite. They all run in the same workspace from one credit balance, so you can run the same prompt on each and compare the results side by side.
No. Vidney runs Wan 2.5 for you, so there are no API keys to manage and no separate provider account to set up.
