SeeBloom
SeeBloom / Models / Wan 3.0
AI Video Model

Wan 3.0 AI Video Generator — Multi-Reference Clips with Native Dialogue

Build longer story clips with multiple character references, native speech and slapstick timing. Ideal when you need several people to stay consistent across a 2–30 second shot.

30 free credits on sign-up · Failed jobs refunded

Wan 3.0 at a glance

What you can set for each generation on SeeBloom.

Clip length
2–30 seconds
Resolution
480p, 720p, 1080p
Aspect ratio
16:9, 9:16, 1:1 and common Studio ratios
Audio
Native dialogue and ambience
Inputs
Text-to-video · multi-image references
Price on SeeBloom
3 / 5 / 10 credits/s (480p / 720p / 1080p)

Why creators choose Wan 3.0

Where the model stands out compared with other video generators.

Multi-character consistency

Attach several reference images and keep each person looking like themselves across the whole clip.

Native dialogue

Write spoken lines in the prompt and get clear on-clip speech without a separate audio pass.

Longer storytelling

Go up to 30 seconds for comedy skits, product stories and social narrative beats.

Resolution tiers

Draft at 480p from 3 credits/s, then finish at 720p or 1080p when you need sharper detail.

Wan 3.0 prompt examples

Copy a prompt, open the studio and paste it in — then adjust the subject and details to make it yours.

Alley fake robbery (multi-ref comedy)
30s · 1080p · 9:16 · multi-ref

Alley fake robbery (multi-ref comedy)

Vertical Pixar-style 3D comedy, 30 seconds. Keep characters matching references: @Image1 chubby little boy in casual T-shirt and shorts with a small crossbody bag (no armor/weapons); @Image2 pretty petite cartoon girl. Night alley with warm street lamps. Slapstick comedy, no gore, no realistic violence. 0-4s: Boy walks in with the bag; girl steps from the shadows and blocks him. 4-8s: Girl plants hands on hips. Girl: "Halt! This is a stick-up!" 8-12s: Boy scoffs, puffs his belly. Boy: "You kidding? With that tiny build, you dare rob me?" 12-16s: She grabs the bag strap; he clings on; they tug back and forth — the bag stays in both their hands. 16-19s: Still tugging, she shouts: "Help! Robbery——!" 19-23s: Two or three bystanders (auntie, uncle) pin him down for cartoon beat-up (slipper taps, back thumps, sitting on his belly). Boy: "Ouch! Wrong guy!" The strap slips; the bag finally ends up with the girl. 23-26s: Girl bolts out of the alley with the bag; bystanders keep pinning the boy. 26-30s: Boy is cartoon-bruised, still pinned; he lifts his head and raises a hand toward her escape, yelling: "Don't run——! Give my bag back——!" Critical: until he is pinned, both must keep tug-of-war on the bag; she only gets it during the beat-up, then flees. Exaggerated faces, clear native dialogue, no on-screen text.
Try it
Prompt idea · multi-image references

Keep two characters on-model

Keep both characters matching the reference images: @Image1 is a chubby boy in a casual tee with a small crossbody bag; @Image2 is a petite cartoon girl. Night alley comedy slapstick with warm street lamps. Clear native dialogue, exaggerated expressions, no on-screen text.
Try it

How to make a Wan 3.0 video

Three steps, no software to install.

1

Add references

Upload the character stills you need to lock identity, then mention them in the prompt.

2

Write the beat sheet

Break the story into timed beats and put dialogue in quotes for clearer speech.

3

Pick length and resolution

Choose 2–30 seconds and 480p / 720p / 1080p, then generate and download the MP4.

Wan 3.0 pricing

Wan 3.0 is billed per second by resolution: about $0.15/s at 480p, $0.25/s at 720p and $0.50/s at 1080p at the base rate (20 credits = $1).

ExampleRateTotal
10s · 480p3 credits/s30 credits
10s · 720p5 credits/s50 credits
30s · 1080p10 credits/s300 credits

Multi-image jobs add 1 credit when image references are used. Failed generations are refunded automatically.

Frequently asked questions

About Wan 3.0 on SeeBloom.

What is Wan 3.0?

Wan 3.0 (Tongyi Wanxiang) is Alibaba’s multimodal video model. On SeeBloom it supports text-to-video and multi-image references with native dialogue.

How many reference images can I use?

Studio supports multiple character / scene references so you can keep several people consistent in one clip.

Does Wan 3.0 generate speech?

Yes. Put spoken lines in quotes and describe the comedy or drama timing you want.

Can I try Wan 3.0 for free?

New accounts get 30 free credits. A 10-second 480p clip costs 30 credits.

Is SeeBloom affiliated with Alibaba?

No. SeeBloom is an independent platform providing access to Wan 3.0 alongside other leading models.

Start creating with Wan 3.0

Sign up to get free credits and generate your first clip in minutes.

Open Wan 3.0 in SeeBloom →