An example of what this page produces.
Seedance 2.5
Seedance 2.5 flagship AI video model
Seedance 2.5 flagship AI video model
Describe the scene, the character action, and the camera move. e.g. “slow dolly-in on a lone figure crossing a neon-lit street in the rain”.
Model
New in 2.5: generate up to 30 seconds, guide motion with as many as 30 images, or start from audio alone.
Free credits are for a quick preview, not the final video. Use a short duration and lower resolution to check the subject, motion, and prompt. A fuller, more compelling final shot often needs a longer duration and higher supported resolution—use paid credits when your scene needs those settings.
Audio
With audio
Synchronized sound, voice and music
Your generated video appears here. Describe a shot and hit Generate.
Start with a short 480p concept preview to check the subject, motion, and prompt direction. When it works, give the action the time it needs — up to 30 seconds — and choose 720p for the fuller Seedance 2.5 take.
Seedance 2.5 runs from 4 to 30 seconds at 480p or 720p, across seven aspect ratios at 24 fps. Reference mode accepts up to 30 images, 10 video clips, and 10 audio files in one request. Use a short 480p run to test the direction. A longer take helps only when the action needs more time, and 720p increases output dimensions — not prompt adherence. Paid credits provide settings and iteration room plus commercial-use eligibility under the Terms; no plan guarantees a perfect result. The maximum reservation is shown before submission, and confirmed failed jobs return it.
Seedance 2.5 is ByteDance's newest Seedance release, built around longer single-pass generation and a far wider reference set.
Earlier Seedance models cap a request at 15 seconds. Seedance 2.5 generates up to 30 seconds in a single pass, so a complete beat can arrive as one take instead of an extend-and-join sequence. Each run is still a separate generation, and one clear action reads better than five plot points.
Reference mode takes up to 30 images, 10 video clips, and 10 audio files together. They guide subject, product, palette, motion, and pacing; they do not lock identity or composition, so review every take.
Seedance 2.5 accepts an audio reference with no image or video attached — the one Seedance tier on FrameAI that does. Supply a track or a voice line, describe the scene, and the generation is guided by the sound you sent rather than fitted to it afterwards.
Seedance 2.5 is ByteDance's newest Seedance video model, positioned around long-form generation: the clip is produced in one pass rather than assembled from shorter windows, which is what makes a 30-second single take possible. FrameAI runs it through ModelArk and enforces the provider's published limits before any credits are reserved.
The practical trade is resolution for length. Seedance 2.5 outputs 480p and 720p; Seedance 2.0 reaches 4K but caps a request at 15 seconds. Choose 2.5 when the shot has to run long, carry many references, or follow supplied audio, and 2.0 when the deliverable has to be high-resolution. Switching models starts a new generation, so results may differ.
A 30-second request is generated as one clip, so you are not joining takes to reach length. It still helps to describe a single continuous action with one camera behaviour rather than a story with several scene changes.
Up to 30 images, 10 clips, and 10 audio files can guide one request — enough for a product from several angles plus a style board. References guide the result; they do not guarantee that packaging, text, or fine detail stay accurate, so review before use.
An audio reference can guide the generation, and generated sound can come back with the picture. The two are separate: a supplied track is an input, while anything returned is generated output to review for timing, content, and copyright restrictions.
Need 1080p or 4K?
Seedance 2.5 runs longer; Seedance 2.0 renders higher, up to 4K, with a 15-second ceiling per request. Reuse the same prompt and references in a new Seedance 2.0 generation when resolution matters more than length. It is a separate model run, so review the new result for differences.
Model specifications
Published model limits, not marketing rounding. Every number below is enforced by the generator before your credits are reserved.
The model is ByteDance's. What FrameAI adds is what makes it usable at work: honest limits, honest billing, and output you are allowed to ship.
Text, up to 30 reference images, 10 reference videos, and 10 audio references can guide one Seedance 2.5 generation instead of being stitched together afterwards.
When audio is enabled, Seedance 2.5 can generate footsteps, room tone, impacts, dialogue, and score with the video. Review timing and content before publishing, or switch audio off when the clip needs separate sound design.
One prompt covers a 9:16 vertical cut for Reels, a 16:9 master for YouTube, and a 21:9 anamorphic frame for a title sequence. Resolution runs from 480p to 4K, so the same take holds up on a cinema screen and on a phone.
The maximum credit reservation is calculated from the selected model, duration, resolution, ratio, and video-reference input, then shown before submission. A successful video job can settle lower from provider-reported usage; confirmed failed jobs return the reservation.
Use Mini for cost-efficient prompt exploration, Fast when you want a balance of generation speed and cost, and Seedance 2.5 when its longer duration or richer reference inputs fit the shot. Switching models starts a new generation, so results may vary.
Content generated with a paid subscription or purchased credit pack may be used commercially under the Terms of Service. Content generated solely with free sign-up or promotional credits is for evaluation and personal use only.
No timeline and no editing software. Describe the shot, attach whatever should guide it, and generate a clip of up to 30 seconds.
Describe subject, action, and camera in English or Chinese. Attach up to 30 images, 10 video clips, or 10 audio files — or send audio on its own. References guide the take rather than locking identity or composition.
Start with a short 480p preview. For a take you want to keep, choose 4 to 30 seconds based on the action, use 720p when output size matters, and leave room for another take. The maximum credit reservation updates before submission; generated audio is on by default and can be switched off.
The job runs in the background and the result waits in My Creations. Output generated with a paid subscription or purchased credit pack may be used commercially under the Terms of Service. If the provider fails the job, the reservation returns.
A longer ceiling changes what fits inside one generation — a full spot, a scene with a beginning and an end, a sound-led cut.
A 30-second ceiling covers a standard social spot without joining clips, so pacing and lighting come from one generation. Attach the product photo to guide appearance; logos, labels, and fine detail can still change, so review every variant before an ad test.
Send a track as an audio reference and describe the scene, and the sound guides the generation instead of being laid over it afterwards. Review the returned timing and content before publishing — a reference guides a new take, it does not synchronize one.
Studios and brand teams can supply a product from several angles, a style board, and a motion clip in one request. The result is guided by all of it, but each run remains a separate generation that should be reviewed for drift.
Seedance 2.5 is ByteDance's newest Seedance video model. On FrameAI it generates 4-to-30-second clips at 480p or 720p and accepts up to 30 reference images, 10 reference clips, and 10 audio files. Seedance 2.0 caps a request at 15 seconds but reaches 4K.
Up to 30 seconds per generation at 24 fps, produced in one pass rather than joined from shorter takes. For longer pieces, Video Extend continues a clip and Video Transition generates a new clip between two endpoint images; both are separate generations that need review.
A text prompt in English or Chinese works on its own. Reference mode adds up to 30 images, 10 video clips, and 10 audio files, and on Seedance 2.5 an audio reference is valid with no image or video attached.
480p and 720p, across 16:9, 9:16, 21:9, 1:1, 4:3, 3:4 and adaptive, which lets the model match the shape of your reference material. The 720p ceiling is the provider's, and the generator enforces it rather than letting a job fail upstream. Use Seedance 2.0 when the deliverable needs 1080p or 4K.
The maximum credit reservation depends on duration, resolution, aspect ratio, and video-reference input, and is displayed before submission. After a successful job, provider-reported usage can lower the final charge and unused credits are returned. Confirmed failed jobs return the reservation.
To the extent permitted by applicable law, FrameAI does not claim ownership of videos you generate. Videos generated with a paid subscription or purchased credit pack may be used commercially under the Terms of Service. Videos generated solely with free sign-up or promotional credits are for evaluation and personal use only.
Describe one continuous action with one camera behaviour, then add light and mood. A 30-second request is still a single shot rather than a story: five plot beats in a paragraph give the model contradictory instructions. If a take drifts, reduce it to one action and add detail back one element at a time.
Yes. When audio generation is enabled, foley, ambience, dialogue, and score come back with the picture. Treat it as generated output: review timing, content, and copyright restrictions, or switch audio off when the clip needs a separate sound-design pass.
No. AI generation is nondeterministic, so no plan can guarantee a perfect take or exact prompt adherence. Paid credits let you use 720p, give the action more time when it needs it, and keep enough budget for another take or prompt refinement. Paid-credit output is also eligible for commercial use under the Terms of Service. Review every result before publishing.
Each tool uses Seedance 2.5 for a different job. Move between them freely — your credits, history, and exports are shared.
Choose paid access when you need commercial-use eligibility, predictable generation costs, and credits restored after a failed or rejected job. One-time packs are available with no subscription.
