Yovid

Wan 2.5 Free Online AI Video Generator, Upgraded to Wan 3.0

Yovid doesn’t run Wan 2.5. It runs Alibaba’s successor, Wan 3.0, which keeps the voices, music and effects that made 2.5 stand out and stretches a clip from 10 seconds to as long as 30. Start from a prompt, a first frame, or up to 10 images and 5 clips. 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

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.

What Wan 3.0 Makes on Yovid

Regency ballroomWan 3.0 · crowd scene, 30 s
Fridge expeditionWan 3.0 · 3D animation, 20 s
Curtain callWan 3.0 · one take, 24 s
Neighborhood caféWan 3.0 · brand film, 22 s
Temple sparringWan 3.0 · animated action, 30 s
Make a Wan 3.0 video

Why Wan 3.0 Takes the Place of Wan 2.5 on Yovid

Yovid Runs Wan 3.0, Not Wan 2.5

Alibaba released Wan 2.5 as a preview on Alibaba Cloud Model Studio in September 2025, as a hosted model rather than downloadable weights like Wan 2.1 and 2.2. Its headline was sound: dialogue, effects and music generated together with the picture, in clips of 5 or 10 seconds at up to 1080p.

Wan 2.6, Wan 2.7 and Wan 3.0 followed within a year. Yovid runs only Wan 3.0, so the composer on this page opens on it and every clip you make here comes from it. The table shows what each step added.

Wan 2.5 · Sep 20255 or 10 s with native sound480p, 720p or 1080p from text or a first frame; prompts up to 1,500 characters.
Wan 2.6 · Dec 2025Up to 15 s, multi-shotAdds reference-to-video that keeps a person’s look and voice consistent.
Wan 2.7 · Apr 2026Up to 15 s, plus video editingUp to five mixed image and video references; prompts up to 5,000 characters.
Wan 3.0 · Aug 2026Up to 30 s, the version on YovidMore lifelike, varied faces and tighter consistency with your references.

Wan 3.0 Runs 30 Seconds, Three Times Wan 2.5

Ten seconds holds one shot. Thirty holds a scene with a setup, a turn and an ending. On the left is a 10-second Wan 2.5 demo: one reporter, one line, one running shot. On the right is a 30-second Wan 3.0 demo built from four reference images, a reunion told across two timelines with a Japanese voice-over.

On Yovid the length is a slider, any whole second from 2 to 30, and the price follows the seconds you choose. Try the idea short, then render the full length once the prompt works.

Wan 2.5 · 10 sWan 3.0 · 30 s

Wan 2.5 Brought Sound into the Same Pass

Sound generated together with the picture was Wan 2.5’s headline feature: a spoken line lands on the speaker’s lips and a footstep on the frame where the foot hits. This 10-second Wan 2.5 Preview demo has a Beijing local greeting his neighbor in a broad Beijing accent while the thrush in his cage chirps on cue.

Wan 3.0 keeps that and gives it more room. On Yovid sound is on by default and costs nothing extra, since a clip is priced the same with sound on or off. Put each spoken line in quotes, say who speaks and in which language, and switch sound off when you plan to score the clip yourself.

Prompt (abridged): Early autumn morning in a Beijing hutong, warm side light on grey flagstones, an old pagoda tree swaying in the breeze. A man in his sixties in a navy cotton jacket and cloth shoes carries a rosewood birdcage; the thrush inside chirps twice. He meets a neighbor, smiles, nods and says in a broad Beijing accent: "哟，吃了么您呐？没事儿遛遛弯儿，舒坦！" (Eaten yet? Just out for a stroll. Feels great!) The cage sways as the bird flutters; pots of jasmine stir by the grey brick wall.

Lip-Synced Dialogue in the Language You Write

Write the line, name the speaker and the language, and Wan 3.0 delivers it with matching lip movement. In this 24-second demo three reference images fix two game characters and a dojo: she asks where they are, he tells her this is her first trial, and a cold system voice starts the test, all in Japanese and each on the right mouth.

Alibaba’s other Wan 3.0 demos carry lines in English, Mandarin, Korean, Portuguese, Thai, Spanish and Arabic. Add a reading direction, such as low and calm or quick and nervous, so the delivery fits the scene.

Lip-Synced Dialogue in the Language You Write – Wan 3.0
Input
Wan 3.0 logoWan 3.0
Prompt (abridged): About 24 s, a cinematic Japanese sci-fi game opening. Image 1 is the silver-haired man in goggles and a long grey coat, Image 2 the woman with twin buns, a red mechanical mask and a white combat suit, Image 3 a wooden dojo. In a blank white virtual space a grid spreads under their feet and the dojo assembles plank by plank. She asks, guarded: 「ここは……何のステージ？」 He raises a palm and answers calmly: 「お前の適性を測る、最初の試練だ。」 The dojo dissolves to white and a system voice says: 「適性試験、開始。」 Accurate Japanese lip sync, no subtitles. (Wan 3.0 · 16:9 · 24s · 1080p)

Up to 10 Images and 5 Clips as References

Wan 2.5 started from a text prompt or a single image used as the first frame. Wan 3.0 on Yovid adds reference mode, where up to 10 images and 5 video clips of 1 to 15 seconds pin down the people, creatures, products and places while the prompt directs them.

This 30-second demo used nine images: one for the woman, one for a lotus realm and seven for small creatures, each drawn in 2D beside its 3D version. Her face holds from start to finish while the creatures step out of their drawings into 3D. Name every file in the prompt (Image 1, Image 2 and so on) and give each one a single job.

Up to 10 Images and 5 Clips as References – Wan 3.0
Input
Wan 3.0 logoWan 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)

Room for a Full Shot List: 10,000-Character Prompts

Wan 2.5 capped prompts at 1,500 characters, about enough for one well-described shot. Wan 3.0 on Yovid accepts up to 10,000, so a whole plan fits: shots by the second, the camera technique for each, the music cues and every spoken line.

This 30-second demo came from one text prompt with no reference images. It laid out nine timed shots, from a dive through the clouds to a slow pull-away, and Wan 3.0 followed the order and the pacing.

Room for a Full Shot List: 10,000-Character Prompts – Wan 3.0
Wan 3.0 logoWan 3.0
Prompt (abridged): A nine-shot, 30-second fantasy in a retro Chinese calendar-poster style: even ink outlines, patterned clouds, watercolor on aged rice paper. 0–4 s: an FPV dive through the clouds past a giant turtle carrying a mountain of palaces. 4–11 s: a first-person ride on a golden qilin low over a festival. 11–19 s: a pipa player builds to one sweeping strum, held in slow motion. 19–21 s: a held breath over the square. 21–27 s: a dancer whirls a flame-orange silk into a ring. 27–30 s: a slow rise as the mountain sinks back into the clouds. Guzheng and pipa over ambient electronics. (Wan 3.0 · 16:9 · 30s · 720p)

Wan 3.0 Settings and Credits on Yovid

Pick the frame before you write so the blocking is composed for it; when you start from an image, adaptive keeps that image’s shape. Credits are charged per second of output, set by resolution alone, and reference clips add their own seconds.

Length2 to 30 seconds5 seconds by default. Wan 2.5 offered 5 or 10.
Resolution480p, 720p or 1080p720p by default; the same three tiers as Wan 2.5.
Aspect ratio16:9, 9:16, 1:1, 4:3, 3:4 or adaptive
InputsText, a first frame with an optional last frame, or referencesUp to 10 images and 5 clips of 1–15 s (MP4 or MOV, up to 100 MB each).
SoundOn by default, no extra creditsVoices, music and effects; switch it off for any generation.
CreditsAbout 1.2, 2.4 or 4.8 per secondAt 480p, 720p or 1080p, credit-pack base rates; a 5-second 720p clip is about 12 credits.

Who Moves from Wan 2.5 to Wan 3.0

Most people looking for Wan 2.5 want a clip that talks and has a soundtrack, without stitching 10-second pieces together. Wan 3.0 covers these jobs best:

  • Short-drama and skit creators: Dialogue scenes that run past 10 seconds, with the same cast held across episodes through reference images.
  • Ad and social teams: Product spots with voice-over and music in one render, the product kept on-model from your own photos.
  • Music and dance video makers: Performances where the lyrics, the beat and the lip sync come out of the same generation.
  • Animators and game studios: Stylized openings and trailers, from ink-and-watercolor to cel-shaded CG, in one 30-second pass.
  • Teams with a Wan 2.5 prompt library: Old prompts are a ready starting point: run one as it is, then add timed beats, references and lines to fill the extra length.

Wan 3.0, Successor to Wan 2.5, vs Seedance 2.5, MiniMax H3 and Kling 3.0

Wan 2.5 is not in the table because Yovid doesn’t run it. Every cell except the last row comes from the settings each model offers in Yovid’s generator today.

Wan 3.0, the successor to Wan 2.5, compared with Seedance 2.5, MiniMax H3 and Kling 3.0 on Yovid
Feature / ModelWan 3.0Seedance 2.5MiniMax H3Kling 3.0
Longest clip30 s30 s15 s15 s
Top resolution1080p1080p2K1080p
Aspect ratios16:9, 9:16, 1:1, 4:3, 3:416:9, 9:16, 1:1, 4:3, 3:4, 21:921:9, 16:9, 4:3, 1:1, 3:4, 9:1616:9, 9:16, 1:1
Reference images10309—
Reference videos5103—
Edit or extend an uploaded clip—Yes——
Credits for 5 s at 720p12312313
Best forThe Wan 2.5 kind of clip, with sound, now up to 30 seconds on a tight budget.Large reference stacks, plus editing or extending uploaded footage.Sharp 2K clips of up to 15 seconds.Lively 15-second character clips at up to 1080p.

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.

Wan 3.0 as a Wan 2.5 Replacement: Strengths and Limits

What you gain by moving to Wan 3.0 on Yovid, and what you give up.

Strengths

  • Three times the length: 2 to 30 seconds per clip, against Wan 2.5’s 5 or 10.
  • Sound at no extra cost: Voices, music and effects stay on by default and cost the same as a silent clip.
  • References Wan 2.5 never had: Up to 10 images and 5 video clips keep people, products and places consistent.
  • Longer prompts: Up to 10,000 characters against Wan 2.5’s 1,500, so a timed shot list fits.

Limits

  • Not Wan 2.5 itself: If a project must match footage already made with Wan 2.5, Yovid can’t reproduce that model; Alibaba Cloud Model Studio still documents the 2.5 preview.
  • No audio upload: Wan 2.5’s API could build a clip around your own audio file; on Yovid, Wan 3.0 works from text, images and video clips.
  • Plan access: Not on the free plan. Lite and above or any one-time credit pack unlocks Wan 3.0, 1080p included.
  • Resolution: Still tops out at 1080p, as Wan 2.5 did. For 2K or 4K, pick MiniMax H3 or Veo 3.1, or run the clip through Yovid’s video upscaler.

How to Run a Wan 2.5-Style Prompt on Wan 3.0

  1. Step 1

    Open Wan 3.0

    Start from this page and the generator opens with Wan 3.0 selected. Wan 2.5 is not in the model list.

  2. Step 2

    Write the scene and the sound

    Describe the shots, quote the dialogue and name its language, then add a first frame or up to 10 images and 5 clips if you have them.

  3. Step 3

    Set the length and generate

    Choose 2 to 30 seconds, a ratio and a resolution, check the credit cost, then generate and download the MP4.

Try Wan 3.0 now

Wan 3.0 Prompts to Start From

Real prompts from Alibaba’s Wan 3.0 showcase, shortened and translated from Chinese, next to the clips they made. Each opens in Yovid’s generator on Wan 3.0; prompts that name Image 1, Image 2 and so on need your own reference images.

  • Curtain call, one take

    Text to video · 24 s

    A 24-second, 16:9 single continuous take with a realistic film look, one 35 mm lens and no cuts. A singer of about 28, dark hair pinned up, in a muted wine-red velvet gown, has nearly missed her entrance. 0–12 s: the camera backs away fast as she threads a narrow backstage, past a makeup artist blotting her forehead, a rolling costume rack and a painted flat pushed across her path; 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 on standby.) 12–21 s: in the wings he raises a hand: "最后八小节。" (Last eight bars.) The orchestra thins to bass and sparse piano, and the camera slowly circles her as 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

  • Eight languages, one hook

    Three references · 24 s

    A 24-second photoreal hip-hop video on a city rooftop at blue hour. Image 1 sets the four-piece crew (female lead vocalist, male DJ, two hype performers), Image 2 the lead’s face and silver-black techwear, Image 3 the rooftop stage and skyline. An original 106 BPM anthem with a two-bar synth-and-brass hook. Eight shots, one language each, every line 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." At the 21-second drop the camera pulls back to the whole crew: "Different voices, one beat." Exact lip sync, the same face and voice throughout, no captions or logos.

    Source: Alibaba Cloud Model Studio · Wan 3.0

    Try this prompt

  • Cat and wolf sparring

    Two references · 30 s

    A 30-second, 16:9, 1080P cinematic 3D animated action short. @Image 1 is the only tabby-cat heroine: brown-grey stripes, amber eyes, teal hooded cape, bronze shoulder guards, ringed tail. @Image 2 is the only grey-wolf hero: layered grey fur, ice-blue eyes, blue-and-gold cloud-pattern robe. Their swords stay sheathed on their backs throughout. In a ruined cave temple under a giant weathered stone Buddha head they spar bare-handed: punches, sweeps and wrist grabs, a chase up broken stairs, a one-second slow-motion clash at the top of a leap, a ledge that gives way until he catches her wrist, and a back-to-back landing in drifting pink petals. No dialogue or music, only impacts, cloth, falling stone, footsteps and wind.

    Source: Alibaba Cloud Model Studio · Wan 3.0

    Try this prompt

  • Into the lotus realm

    Nine references · 30 s

    A 30-second, 16:9 cinematic surreal Eastern fantasy. Image 1 is the only identity reference for a young woman with long, softly wavy dark-brown hair, an open light-blue denim shirt and cream trousers; keep her face identical in every shot. Image 2 fixes a hidden realm of giant translucent lotus flowers glowing peach, coral and pale gold. Images 3–9 each show one small creature, 2D on the left and 3D on the right. She parts the vines and walks in; she touches a lotus and a glowing drop reflects three of the 2D creatures; they leap a stream and turn into their 3D forms; she kneels, smiling, as they gather around her; the camera rises as every lotus blooms at once. Live action blended with film-grade CG, no captions or logos.

    Source: Alibaba Cloud Model Studio · Wan 3.0

    Try this prompt
Read the prompt-writing guide

More Video Models on Yovid

Wan 3.0Seedance 2.5MiniMax H3Kling 3.0Veo 3.1Hailuo 2.3Sora 2

Guides for Moving from Wan 2.5 to Wan 3.0

Compare every video model on Yovid, then write the longer, timed prompts that make use of Wan 3.0’s extra length.

All 14 Yovid video models compared, Wan 3.0 includedA shot-by-shot formula for text-to-video prompts
Image-to-video prompts: six examples taken apart
Browse all Yovid guides

FAQs

What is Wan 2.5?

Wan 2.5 is a video generation model from Alibaba’s Wan team, released as a preview on Alibaba Cloud Model Studio in September 2025. It made 5- or 10-second clips at 480p, 720p or 1080p from text or a single image, with dialogue, effects and music generated together with the picture.

Can I use Wan 2.5 free online on Yovid?

No. Yovid doesn’t run Wan 2.5. It runs the successor, Wan 3.0, which keeps the native sound and makes clips of up to 30 seconds. Signing up for Yovid is free and includes credits for the free-plan models; Wan 3.0 itself needs the Lite plan or above, or any one-time credit pack.

Is Wan 2.5 open source?

No. Alibaba published open weights for Wan 2.1 and Wan 2.2, but Wan 2.5 came out as a hosted preview that you use through Alibaba’s services rather than download.

What changed from Wan 2.5 to Wan 3.0?

The longest clip went from 10 seconds to 30. Along the way Wan 2.6 added multi-shot scenes and video references, Wan 2.7 added video editing, and Wan 3.0 brought more lifelike, varied faces and tighter consistency with your references. Generated sound stayed throughout.

How long can a Wan 3.0 video be on Yovid?

Any whole number of seconds from 2 to 30 per generation, as one continuous clip; 5 seconds is the default. That is three times Wan 2.5’s 10-second maximum.

Does Wan 3.0 generate audio like Wan 2.5?

Yes. Sound is on by default and covers voices, effects and music, with lip sync on spoken lines. It costs no extra credits and you can switch it off for any clip. Unlike Wan 2.5’s API, Yovid doesn’t take your own audio file as an input.

How many credits does Wan 3.0 use?

At credit-pack base rates, 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 full 30-second 1080p clip about 144. Reference clips add their own seconds to the bill.

Can I reuse my Wan 2.5 prompts?

Yes, as a starting point. Wan 2.5 prompts were capped at 1,500 characters and Wan 3.0 on Yovid accepts up to 10,000, so any of them fits. Expect a different result, since it is a different model, and use the extra room for timed beats, quoted dialogue and references.

Which resolutions and aspect ratios does Wan 3.0 support?

480p, 720p and 1080p, the same tiers Wan 2.5 offered, in 16:9, 9:16, 1:1, 4:3 or 3:4, or adaptive to follow your first frame. 1080p is included on every paid plan and credit pack.

Is Yovid affiliated with Alibaba or the Wan team?

No. Yovid is an independent platform that runs Wan 3.0 alongside video models from ByteDance, Kuaishou, Google, MiniMax and OpenAI. It is not operated or endorsed by Alibaba.

Make the Clip Wan 2.5 Couldn’t: 30 Seconds, Sound Included

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

Create with Wan 3.0