Seedance 2.5 AI Video Generator

30 seconds in one take. Up to 50 references. Native audio. Start from text, frames, or a reference pack, then confirm 4–30 seconds and 480p–1080p before you generate.

From 38 credits
4–30s · 480p–1080p
Text, frames, or references
ByteDance · Seedance 2.5 docs ↗

Seedance 2.5 Video Examples

Lantern-festival street story used as a Seedance 2.5 example

Lantern festival story

Prompt

Night aerial over a sprawling ancient Chinese city. Warm lantern grids trace tiled roofs and a winding river; golden fireworks burst above distant mountains. Cut to street level: a scholar in a dark teal robe and black futou walks into a lantern-lined crowd past women in pale hanfu and a large colorful parade lantern. End on a medium close-up of the scholar as lantern bokeh fills the night. Keep the festival glow and crowd density locked.

Key Features of Seedance 2.5

30-second storytelling with precise reference control—each row has a matching demo. Prefill the brief and any references; nothing runs until you generate.

30-second storytelling

Tell a complete beat in one take. Seedance 2.5 holds setup, turn, and finish across a continuous 4–30 second clip—without stitching shorter takes.

One-take backstage-to-stage concert story generated with Seedance 2.5

How to Use Seedance 2.5

1
1. Start from text, frames, or references
Use text-to-video for a new scene, a first frame with an optional last frame, or multimodal for up to 30 images, 10 videos, and 10 audio—50 files total. First/last-frame and pack modes cannot be combined. Prefill from a demo if you want—nothing runs until you generate.
Use text-to-video for a new scene, a first frame with an optional last frame, or multimodal for up to 30 images, 10 videos, and 10 audio—50 files total. First/last-frame and pack modes cannot be combined. Prefill from a demo if you want—nothing runs until you generate.
2
2. Direct motion, camera, and sound
Write subject, action, setting, camera, and sound. For a 4–30 second take, cover setup, turn, and finish—not a single frozen moment. Tag each file with @ImageN, @VideoN, or @AudioN and say what it provides; vague “use the references” often mixes roles.
Write subject, action, setting, camera, and sound. For a 4–30 second take, cover setup, turn, and finish—not a single frozen moment. Tag each file with @ImageN, @VideoN, or @AudioN and say what it provides; vague “use the references” often mixes roles.
3
3. Generate, review, then iterate
Select 4–30 seconds and 480p, 720p, or 1080p, confirm the credit estimate, then generate. Review identity, logos, dialogue, and lip sync before publishing or iterating.
Select 4–30 seconds and 480p, 720p, or 1080p, confirm the credit estimate, then generate. Review identity, logos, dialogue, and lip sync before publishing or iterating.

Create with Seedance 2.5 ↑

Seedance 2.5 Pricing

From 38 credits per 4-second 480p clip, up to 1,320 with a full reference-video request. Duration, 480p–1080p, and reference-video seconds change the quote. Confirm the exact reservation before you generate.

Compare Seedance 2.5How Seedance 2.5 compares to other models

Pick by the job: one-take length, reference-pack size, resolution, or an instruction-led edit. Specs below are the live MixVio routes.

Seedance 2.5

This model

4–30s one-takes, packs up to 50 files, and clay-render blocking

4–30s38–1320 credits

Wan 3.0 →Alibaba

2–30s one-takes, 10/5/5 reference packs, and document or webpage input

2–30s12–660 credits

Grok Imagine Video 1.5 →xAI

1–15s videos with native audio, a start photo, or up to 7 photo refs

1–15s6–264 credits

Attribute	This modelSeedance 2.5	AlibabaWan 3.0 →	xAIGrok Imagine Video 1.5 →	MiniMaxMiniMax H3 →
Best for	4–30s one-takes, packs up to 50 files, and clay-render blocking	2–30s one-takes, 10/5/5 reference packs, and document or webpage input	1–15s videos with native audio, a start photo, or up to 7 photo refs	Multimodal reference, first/last frames, and instruction-led edits
Inputs	Text · Up to 30 images · 10 videos · 10 audio	Text · Up to 10 images · 5 videos · 5 audio	Text · Start photo · Up to 7 photos	Text · First/last frame · Up to 9 images · 3 videos · 3 audio
Output	480p–1080p video + audio	480p–1080p video + audio	480p–1080p video + audio	768p–2K video + audio
Duration	4–30s	2–30s	1–15s	4–15s
Native audio	Yes	Yes	Yes	Yes
Credits	38–1320 credits	12–660 credits	6–264 credits	12–128 credits

Stay on Seedance 2.5 for a 4–30 second one-take, packs up to 50 files, or clay-render blocking. Switch to Wan 3.0 for a 2–30 second take with a 10/5/5 pack or a public document or webpage, Grok Imagine Video 1.5 for 1–15s photo-led takes, MiniMax H3 for 4–15s instruction-led edits at 768p or 2K, or PixVerse V6 for 1–15 second videos from text or one photo with optional audio and multi-shot.

Seedance 2.5Wan 3.0MiniMax H3Grok Imagine Video 1.5

Seedance 2.5 FAQs

What is Seedance 2.5?

Seedance 2.5 is ByteDance Seed’s next video model for 4–30 second audio-video clips. On MixVio it creates one private clip per run from text, a first frame with an optional last frame, or a multimodal pack of images, videos, and audio. Seedance 2.5 announcement ↗

What is Seedance 2.5 best for?

Use it for a 4–30 second one-take, a large owned reference pack, clay-render blocking, or a product and character lock with native dialogue. Write the brief like a shot list—subject, action, setting, camera, and sound—and cover setup, turn, and finish on a longer take.

What inputs and file limits does Seedance 2.5 support?

Use Text to Video for a scene brief, Image to Video for one owned start frame with an optional end frame, or Multi Reference for owned packs: up to 30 images (≤30 MB each), 10 videos (MP4/MOV, ≤200 MB each), and 10 audio clips (MP3/WAV, ≤15 MB each)—up to 50 files total. Still images need 300–6,000 px on each side and a 2:5–5:2 crop; oversized photos are fitted automatically. Reference videos need 300–6,000 px per side, a 2:5–5:2 aspect ratio, 409,600–8,295,044 total pixels, and 24–60 FPS. Each reference video or audio clip is 2–30 seconds; each media type may total 30 seconds. Audio-only reference packs are supported. First/last-frame mode cannot be combined with a multimodal pack.

How do I tag references with @ImageN, @VideoN, and @AudioN?

After you upload files, mention each one by index in the prompt—@Image1, @Video1, @Audio1, and so on—and give it one job. Example: “Use @Image1 as the look. Match the camera in @Video1. Follow the beat of @Audio1.” Numbering follows upload order. Vague lines like “use the references” often mix roles.

Can I edit or extend an existing video with Seedance 2.5?

You can upload an owned clip in Multi Reference, tag it with @VideoN, and describe restaging, a camera change, or what should happen next. MixVio still returns one new private 4–30 second video per run. There is no separate public extend tool, in-place timeline editor, or multi-round session on the current MixVio routes.

Does Seedance 2.5 generate native audio?

Yes. Public Seedance 2.5 routes return video with synchronized native audio. Describe dialogue, ambience, sound effects, or music in the brief, then review the mix, speech, and lip sync before publishing. Audio on or off does not change the credit quote.

Which ratios, resolutions, and durations are supported?

Text-to-video supports Auto, 21:9, 16:9, 4:3, 1:1, 3:4, and 9:16. First-frame image-to-video follows the source. Each run produces one 4–30 second clip at 480p, 720p, or 1080p. MixVio does not publish 4K on Seedance 2.5. Duration is the length you select—it is never changed to auto without confirmation.

How do I keep character or product consistency?

Upload clear owned references of the subject, tag them with @ImageN (and @VideoN when motion must match), and name the same identity, wardrobe, and product details in the prompt. Use the smallest pack that defines identity, wardrobe, scene, motion, and sound. Fine details can still drift on long action or many scene changes—review faces, logos, and labels before publishing.

How many credits does Seedance 2.5 use?

Seedance 2.5 is available on the Free plan when your balance covers the quote. Runs reserve from 38 credits. Typical anchors: 38 credits for 4 seconds at 480p, 615 for 30 seconds at 720p, and 1,095 for 30 seconds at 1080p. Reference-led requests that include video input can reach 1,320 credits because input and output seconds are billed together. The generator shows the exact reservation before submission. Failed runs cost no credits.

How does Seedance 2.5 compare with Wan 3.0, Grok Imagine Video 1.5, and MiniMax H3?

Stay on Seedance 2.5 for a one-take lasting 4–30 seconds, 480p–1080p, a pack of up to 30 images / 10 videos / 10 audio, or clay-render blocking. Wan 3.0 covers 2–30 seconds with a 10/5/5 pack or a public document or webpage. Grok Imagine Video 1.5 covers 1–15 second photo-led takes with native audio and up to 7 photo refs. Choose MiniMax H3 for 4–15 seconds of instruction-led edits at 768p or 2K. Compare live credits, inputs, duration, and audio before generating.

Can I use Seedance 2.5 videos commercially?

Paid-plan outputs may be used commercially, subject to the Terms of Service and current ByteDance terms. You must have rights to every prompt, reference, likeness, brand asset, voice, and sound direction. Review identity, logos, dialogue, lip sync, consent, and audio before publishing. MixVio does not guarantee exclusivity or trademark clearance.

Can I use real faces or celebrity likenesses?

Only use faces, voices, and likenesses you own or are authorized to use. Public figures, deceptive impersonation, deepfake-style face swaps, minors, and other prohibited content can be blocked by MixVio moderation or Provider safety checks. When a request is allowed, still review identity and consent before any commercial or public use.

Direct a 30-second one-take with Seedance 2.5.
Private. Credits shown first.

Start from text, frames, or an owned pack—including clay-render blocking. Confirm 4–30 seconds and 480p–1080p, then generate one private clip with native audio.