Wan 3.0 Free Online AI Video Generator

Write a director-style prompt, start from a first frame, or stack up to 10 images and 5 clips as references, and Wan 3.0 returns one 2–30 second clip with voices, music and effects already mixed in. Render at 480p, 720p or 1080p. Start creating with Wan 3.0 on Yovid.

Seedance 2.5 is live

Clips up to 30 seconds with image, video, and audio references.

Kling 3.0

Kling's latest video model, ready in your workspace.

Veo 3.1

Google's video model with native audio.

GPT Image 2.5

OpenAI's image model for creating and editing images.

AI Image Upscaler

Sharpen and enlarge images for cleaner exports.

AI Canvas

Chain image and video steps on one visual board.

See What Wan 3.0 Can Make

Train reunionTwo timelines, 30 s

Regency ballCrowd scene, 30 s

Fridge expedition3D animation, 20 s

Café brand filmWarm product story

Dojo loadingJapanese dialogue, 24 s

Make a Wan 3.0 video

Why Choose Wan 3.0 on Yovid

A Whole 30-Second Story in One Generation: Set any length from 2 to 30 seconds and get an opening, a build and an ending in one clip.
One Continuous Take, Directed by the Prompt: Lay out the camera route and tempo beat by beat, and Wan 3.0 keeps a single unbroken shot.
Voices, Music and Effects in the Same Pass: Dialogue, songs and sound effects are generated with the picture, lip-synced, in many languages.
Up to 10 Images and 5 Clips as References: Pin people, creatures, products and places with a reference stack, then direct them in words.
Characters That Stay On-Model at Full Speed: Fur, costumes and props hold through 30 seconds of chases, falls and close combat.
Every Wan 3.0 Setting in One Place: Five aspect ratios plus adaptive, three resolutions, 2–30 seconds and three ways to start.
Low Cost per Second, Sound Included: About 2.4 credits a second at 720p, with the same price whether sound is on or off.

A Whole 30-Second Story in One Generation

Wan 3.0 runs twice as long as the 15 seconds of Wan 2.7, which leaves room for an establishing shot, a climb in energy, a peak and a quiet close. The clip below runs the full 30 seconds from a single text prompt, from a dive through the clouds to a festival and a slow pull-away, with no reference images at all.

On Yovid you pick the length on a slider, any whole number of seconds from 2 to 30. Work the idea out at 5 seconds, then spend the full 30 on the version you mean to keep, since every extra second is billed.

A Whole 30-Second Story in One Generation – Wan 3.0 — Prompt (abridged): Nine shots, 30 seconds, in a retro Chinese calendar-poster style with even ink outlines, patterned clouds and aged rice-paper watercolor. A giant turtle carries a mountain of palaces through a sea of clouds; a fairy rides a golden qilin down over a festival, a pipa player strikes a sweeping chord, a dancer whirls a long red silk, and the square peaks before the mountain drifts off into the clouds. (Wan 3.0 · 16:9 · 30s · 720p)

One Continuous Take, Directed by the Prompt

Length only helps if the camera behaves. Here the prompt maps a route through a theatre (dressing corridor, costume racks, the wings, the stage) and a tempo for each stretch, and the model holds one 24-second shot: a sprint through the crew, a sudden stillness in the wings, then the curtain lifting on a full house.

Timed beats such as "0–7 s" and "12–14 s" give it the clearest instructions. Wan 3.0 on Yovid accepts prompts of up to 10,000 characters, so a complete shot list with dialogue fits.

One Continuous Take, Directed by the Prompt – Wan 3.0 — Prompt (abridged): A 24-second single take, realistic film look, one 35 mm lens and no cuts. A singer in a wine-red velvet gown nearly misses her entrance: 0–12 s the camera backs away fast as she dodges racks and a sliding set wall; 12–21 s she stops in the wings, the stage manager whispers "最后八小节。" (Last eight bars.), the music thins and the camera slowly circles her; 21–24 s the cue light turns green, the curtain rises and she walks into the light. (Wan 3.0 · 16:9 · 24s · 1080p)

Voices, Music and Effects in the Same Pass

Sound is on by default. Wan 3.0 builds the soundtrack together with the frames, so each line lands on the right mouth, each footstep on the beat, and the music turns with the cut. In this demo one vocalist raps the same hook in eight languages, including English, Mandarin, Japanese, Korean, Thai and Arabic, with the same face, outfit and voice throughout.

Put spoken lines in quotes and say who speaks them and in which language. Audio is free on Wan 3.0: the credit price is the same with sound on or off.

Voices, Music and Effects in the Same Pass – Wan 3.0 — Prompt (abridged): A 24-second hip-hop video on a city rooftop at blue hour. Image 1 fixes the four-person crew, Image 2 the lead vocalist, Image 3 the rooftop stage. One catchy 106 BPM hook; the lead switches language on the beat: "One beat." "同一个节拍。" "ひとつのリズム。" "하나의 리듬." "Um só ritmo." "จังหวะเดียวกัน." "Un mismo ritmo." "إيقاع واحد." Ending: "Different voices, one beat." Precise lip sync. (Wan 3.0 · 16:9 · 24s · 1080p)

Prompt (abridged): A 24-second hip-hop video on a city rooftop at blue hour. Image 1 fixes the four-person crew, Image 2 the lead vocalist, Image 3 the rooftop stage. One catchy 106 BPM hook; the lead switches language on the beat: "One beat." "同一个节拍。" "ひとつのリズム。" "하나의 리듬." "Um só ritmo." "จังหวะเดียวกัน." "Un mismo ritmo." "إيقاع واحد." Ending: "Different voices, one beat." Precise lip sync. (Wan 3.0 · 16:9 · 24s · 1080p)

Up to 10 Images and 5 Clips as References

Switch the composer to references and Wan 3.0 takes up to 10 images and 5 video clips of 1 to 15 seconds in one request. Give every file one job and name it in the prompt: @Image 1 for the person, @Image 2 for the setting, the rest for characters or props.

This demo used nine images: one young woman, one lotus-garden setting, and seven creatures each drawn in 2D and modelled in 3D. Her face holds from the first frame to the last while the creatures step out of their flat drawings into 3D.

Up to 10 Images and 5 Clips as References – Wan 3.0 — Prompt (abridged): 30 s, 16:9, surreal Eastern fantasy. Image 1 is the only identity reference for a young woman in an open light-blue denim shirt; Image 2 fixes a hidden realm of giant translucent lotus flowers; Images 3–9 each show one creature in 2D (left) and 3D (right). She parts the vines, enters the realm and touches a lotus; a glowing drop reveals the 2D creatures, which turn into their 3D forms and gather around her. (Wan 3.0 · 16:9 · 30s · 1080p)

Prompt (abridged): 30 s, 16:9, surreal Eastern fantasy. Image 1 is the only identity reference for a young woman in an open light-blue denim shirt; Image 2 fixes a hidden realm of giant translucent lotus flowers; Images 3–9 each show one creature in 2D (left) and 3D (right). She parts the vines, enters the realm and touches a lotus; a glowing drop reveals the 2D creatures, which turn into their 3D forms and gather around her. (Wan 3.0 · 16:9 · 30s · 1080p)

Characters That Stay On-Model at Full Speed

Fast action is where generated characters usually drift. In this two-image demo a pair of animal martial artists spar through a ruined cave temple for 30 seconds, with a stair chase, a slow-motion clash at the top of a jump and a collapsing ledge, and their fur markings, robes, tails and sheathed swords stay as designed.

The prompt carries most of the weight: it lists what must not change for each reference, splits the fight into timed beats and names every sound, from impacts and cloth to falling stone.

Prompt (abridged): 30 s, 16:9, cinematic 3D animated action. @Image 1 is the only tabby-cat heroine (teal hooded cape, bronze armor, ringed tail); @Image 2 is the only male hero (blue-and-gold cloud robe). Swords stay sheathed on their backs. They spar hand to hand in a ruined cave temple: a chase up broken stairs, a one-second slow-motion clash mid-air, a collapsing ledge, a back-to-back landing in drifting petals.

Every Wan 3.0 Setting in One Place

Choose the frame before you write, so the blocking is composed for it. When you start from an image, Wan 3.0 keeps that image’s shape instead of forcing a preset ratio.

Length	2 to 30 seconds5 seconds by default.
Resolution	480p, 720p or 1080p720p by default.
Aspect ratio	16:9, 9:16, 1:1, 4:3, 3:4 or adaptive
Inputs	Text, a first frame with an optional last frame, or referencesReferences: up to 10 images and 5 clips of 1–15 s (MP4 or MOV, up to 100 MB each).
Sound	On by defaultVoices, music and effects; switch it off for any generation.
Prompt length	Up to 10,000 characters
Output	MP4

Low Cost per Second, Sound Included

Wan 3.0 is billed per second of output, and only the resolution changes the rate. Sound, ratio and input type don’t. When you add reference clips, their seconds count too, billed as input plus output.

That keeps long clips within reach: settle the prompt at 480p, then run it again at 1080p for the version you publish.

480p	About 1.2 credits per secondA full 30-second clip is about 36 credits.
720p	About 2.4 credits per secondThe default; a 5-second clip is about 12 credits.
1080p	About 4.8 credits per secondA full 30-second clip is about 144 credits.
Reference clips	Billed as input plus output secondsA 10-second reference with a 10-second result is billed as 20 seconds.

Who Wan 3.0 Is Built For

Thirty seconds with sound, a ten-image reference stack and a low per-second price suit these jobs best:

Short-drama and series creators: Half-minute scenes with spoken lines, and the same cast held across episodes by reference images.
Advertisers and brand teams: Product spots and brand films with voice-over and music, the product kept on-model from your own photos.
Music and dance video makers: Performances where lip sync, beat and cuts come out of one generation, in any of several languages.
Animators and game studios: Stylized openings and trailers, from ink-and-watercolor to cel-shaded CG, in a single 30-second pass.
Venue and travel marketers: Mood films for cafés, hotels and destinations, built around the photos you already have.

Wan 3.0 vs Seedance 2.5, MiniMax H3, Kling 3.0 and Veo 3.1

Every cell except the last row comes from the settings each model offers in Yovid’s generator today, so you can weigh length, references and cost before you spend a credit.

Wan 3.0 compared with Seedance 2.5, MiniMax H3, Kling 3.0 and Veo 3.1 Fast on Yovid
Feature / Model	Wan 3.0	Seedance 2.5	MiniMax H3	Kling 3.0	Veo 3.1 Fast
Longest clip	30 s	30 s	15 s	15 s	8 s
Top resolution	1080p	1080p	2K	1080p	4K
Aspect ratios	16:9, 9:16, 1:1, 4:3, 3:4	16:9, 9:16, 1:1, 4:3, 3:4, 21:9	21:9, 16:9, 4:3, 1:1, 3:4, 9:16	16:9, 9:16, 1:1	16:9, 9:16
Reference images	10	30	9	—	—
Reference videos	5	10	3	—	—
Edit or extend an uploaded clip	—	Yes	—	—	—
Credits for 5 s at 720p	12	31	23	13	15
Best for	30-second clips with sound at the lowest price per second.	The biggest reference stacks, plus editing or extending footage.	2K clips up to 15 seconds from a few references.	Lively 15-second character clips at 1080p.	Eight-second hero shots, up to 4K.

Credits are base credit-pack rates for a 5-second clip at 720p (or the closest resolution a model offers) and follow Yovid’s live pricing.

Strengths and Limits of Wan 3.0

Where Wan 3.0 is the best pick on Yovid, and where a different model or tool fits better.

Strengths

30 seconds with sound: Twice the length of Wan 2.7, with dialogue, music and effects at no extra credit cost.
Price per second: About 2.4 credits a second at 720p, less than half the rate of Seedance 2.5.
Reference stack: Up to 10 images and 5 clips in one request keep people, products and places consistent.
Long prompts: Accepts up to 10,000 characters, so a timed shot list with every line of dialogue fits.

Limits

Resolution: Stops at 1080p. For 2K or 4K, pick MiniMax H3 or Veo 3.1, or run the clip through Yovid’s video upscaler.
Plan access: Not on the free plan. Any paid plan or one-time credit pack unlocks it, 1080p included.
Fewer inputs than the API: Alibaba’s own API also takes audio, documents and web pages and can edit or extend clips; on Yovid you work from text, images and video clips.
On-screen text: Alibaba notes that text rendering and audio texture are still improving, so add titles and captions in an editor.

Make a Wan 3.0 Video in Three Steps

Step 1
Open Wan 3.0
Start from this page, or open the video generator and pick Wan 3.0 from the model list.
Step 2
Describe the scene, add inputs
Write the shots, sound and dialogue, add a first frame or up to 10 images and 5 clips, then set length, ratio and resolution.
Step 3
Generate and download
Watch the result with sound on, tighten the prompt if a beat is off, and download the MP4.

Try Wan 3.0 now

Wan 3.0 Prompts Behind the Demos

Real Wan 3.0 prompts, shortened and translated from Chinese, next to the clips they made. Open one in the generator and swap in your own details; prompts that mention Image 1, 2 and so on need your own reference images.

The last eight bars
Text to video · 24 s
A 24-second single take, 16:9, realistic film look, one 35 mm lens, no cuts. A singer of about 28, dark hair pinned up, in a deep wine-red velvet gown, nearly misses her entrance. 0–12 s: the camera backs away fast as she weaves through a narrow backstage past a costume rack, a sliding painted flat and a prop cart; the stage manager says into his headset, "布景就位，灯光准备，主幕待命。" (Set in place, lights ready, main curtain standing by.) 12–21 s: in the wings he raises a hand: "最后八小节。" (Last eight bars.) The music thins, the camera slowly circles her and she closes her eyes. 21–24 s: "Three… two… one!" The cue light turns green, the curtain rises and she steps into the light before a full house.
Source: Alibaba Cloud Model Studio · Wan 3.0
Try this prompt
Feast above the clouds
Text to video · 30 s
A nine-shot, 30-second Eastern fantasy in a retro Chinese calendar-poster style: even ink outlines, patterned cotton-ball clouds, aged rice-paper watercolor in teal and peach. 0–4 s: an FPV dive from the sky past a giant turtle carrying a mountain of palaces. 4–11 s: riding a golden qilin low and fast over a festival of musicians and dancers. 11–19 s: a pipa player builds to one sweeping chord. 21–27 s: a dancer whirls a long red silk into a ring of flame. 27–30 s: a slow pull-up as the mountain drifts into the clouds. Guzheng, pipa and bells over soft electronics.
Source: Alibaba Cloud Model Studio · Wan 3.0
Try this prompt
Dojo loading
Three references · 24 s
About 24 s, a cinematic Japanese sci-fi game opening. Image 1 is the male character (silver hair, goggles, long grey coat), Image 2 the female character (twin buns, red mechanical mask, white combat suit), Image 3 a traditional wooden dojo. In a blank white virtual space, grid lines spread under their feet and the dojo builds itself plank by plank. She looks around and asks, 「ここは……何のステージ？」 He raises a palm: 「お前の適性を測る、最初の試練だ。」 The dojo dissolves back to white and a system voice says 「適性試験、開始。」 Accurate Japanese lip sync, no subtitles.
Source: Alibaba Cloud Model Studio · Wan 3.0
Try this prompt
One beat, eight languages
Three references · 24 s
A 24-second international hip-hop video, photoreal, rooftop at blue hour. Image 1: the four-person crew (lead vocalist, DJ, two hype performers). Image 2: the lead vocalist. Image 3: the rooftop stage and skyline. A catchy 106 BPM anthem with a two-bar synth and brass hook. Eight shots, one language each, landing on the beat: "One beat." "同一个节拍。" "ひとつのリズム。" "하나의 리듬." "Um só ritmo." "จังหวะเดียวกัน." "Un mismo ritmo." "إيقاع واحد." Beat drop at 21 s: "Different voices, one beat." Precise lip sync, same face and voice throughout.
Source: Alibaba Cloud Model Studio · Wan 3.0
Try this prompt
Ruins sparring
Two references · 30 s
A 30-second, 16:9, 1080P cinematic 3D animated action short. @Image 1 is the only tabby-cat heroine: teal hooded cape, bronze shoulder armor, ringed tail. @Image 2 is the only male hero: blue-and-gold cloud-pattern robe, blue bracers. Their swords stay sheathed on their backs. In a ruined cave temple with a giant stone Buddha head, they spar hand to hand: punches and sweeps, a chase up broken stairs, a one-second slow-motion clash mid-air, a collapsing ledge where he catches her wrist, and a back-to-back landing in drifting pink petals. No dialogue or music; only impacts, cloth, falling stone and wind.
Source: Alibaba Cloud Model Studio · Wan 3.0
Try this prompt
Lotus realm
Nine references · 30 s
A 30-second, 16:9 surreal Eastern fantasy. Image 1 is the only identity reference for a young woman with long wavy brown hair, an open light-blue denim shirt and cream trousers. Image 2 fixes a hidden realm of giant translucent lotus flowers. Images 3–9 each show one creature in 2D (left) and 3D (right). She parts the vines and enters the realm; she touches a lotus and a glowing drop reflects the 2D creatures; they leap across a stream and turn into their 3D forms; she kneels, smiling, as they gather; the camera rises as every lotus blooms at once.
Source: Alibaba Cloud Model Studio · Wan 3.0
Try this prompt

Read the prompt-writing guide

Seedance 2.5 Wan 2.5 MiniMax H3 Kling 3.0 Veo 3.1 Seedance 2.0 Sora 2

Guides for Better Wan 3.0 Videos

How to structure a long prompt, image-to-video examples and a side-by-side comparison of every video model on Yovid.

Writing a text-to-video prompt, shot by shot Six image-to-video prompt examples, broken down

AI video models compared on length, resolution and references

Browse all Yovid guides

FAQs

What is Wan 3.0?

Wan 3.0 is the newest video model in Alibaba’s Wan family, released on Alibaba Cloud Model Studio in August 2026. It makes clips of up to 30 seconds with generated sound from text, images and video references, with a focus on lifelike faces and consistent characters. Yovid runs it in the browser alongside Seedance, Kling, Veo and other models.

Can I use Wan 3.0 for free?

Creating a Yovid account is free and comes with sign-up credits for the free-plan models. Wan 3.0 itself needs a paid plan (Lite and up) or any one-time credit pack. At about 1.2 credits a second at 480p, testing an idea costs little.

What is the maximum length of a Wan 3.0 video?

Any whole number of seconds from 2 to 30 per generation, as one continuous clip. The default is 5 seconds.

How many credits does Wan 3.0 use?

At credit-pack base rates it is about 1.2 credits per second at 480p, 2.4 at 720p and 4.8 at 1080p, so a 5-second 720p clip is about 12 credits and a 30-second 1080p clip about 144. Sound costs nothing extra. Reference clips are billed as input plus output seconds.

Does Wan 3.0 make audio and lip-synced dialogue?

Yes. Sound is on by default and covers voices, music and effects, with lip sync on spoken lines. Put dialogue in quotes and name the language; Alibaba’s demos include lines in English, Mandarin, Japanese, Korean, Portuguese, Thai, Spanish and Arabic. You can turn sound off for any generation.

What can I upload as a reference?

Either a first frame with an optional last frame, or a reference stack of up to 10 images and 5 video clips (1–15 seconds each, MP4 or MOV, up to 100 MB). Refer to each file in the prompt, for example @Image 1 or @Video 1. Audio, document and web-page inputs from Alibaba’s API are not available on Yovid.

Which resolutions and aspect ratios does Wan 3.0 support?

480p, 720p and 1080p, in 16:9, 9:16, 1:1, 4:3 or 3:4, or adaptive to follow your first frame. 1080p is available on every paid plan and credit pack.

Wan 3.0 vs Wan 2.7 and Wan 2.5: what changed?

Wan 3.0 doubles Wan 2.7’s 15-second limit to 30 seconds, renders more natural, varied faces, and holds references more tightly. Yovid runs Wan 3.0; Wan 2.5 and Wan 2.7 are not offered here.

Wan 3.0 or Seedance 2.5: which should I use?

Both reach 30 seconds with sound. Wan 3.0 costs about 2.4 credits a second at 720p against roughly 6 for Seedance 2.5. Seedance 2.5 takes larger reference stacks (30 images and 10 clips) and can edit or extend uploaded footage, which Wan 3.0 cannot do on Yovid.

Can I use Wan 3.0 videos commercially?

Commercial use is included on Pro, Premium, yearly plans and one-time credit packs; Lite monthly does not include it. You remain responsible for the rights to your prompts and references, including faces, brands and music.

Your 30-Second Scene, Soundtrack Included

Write the shots, add references if you have them, and let Wan 3.0 render the picture and the sound together.

Create with Wan 3.0