Seedance 2.5 is ByteDance's next-generation AI video model, released July 31, 2026, that generates up to 30 seconds of synchronized audio-video in a single continuous pass — double Seedance 2.0's native duration — with up to 50 multimodal reference inputs and timestamp-level regional editing.
Built for video creators, marketers, and studios, it turns briefs into longer, more consistent sequences with director-style camera control and native audio in 10+ languages, without stitching multiple generations together the way most current video models require.
Long-form directed sequences, multi-subject scenes, product demos, localized video
Video and marketing teams that need longer single-shot sequences without stitching multiple generations together
Double the duration, four times the references — the longest single-pass generation in ByteDance's video line.
Up to 30 seconds of video in a single continuous pass — double Seedance 2.0's native ceiling — with multi-round extension for anything longer. ByteDance calls this "Long-Form Shot Consistency": steadier characters and more complex sequences held across the full clip without stitching generations together.
Load up to 50 reference items in one generation — 30 images, 10 video clips, and 10 audio clips — more than four times Seedance 2.0's 12-reference budget. That's enough headroom for a full brand style guide, a character turnaround, and reference audio in the same request.
Adjust a specific moment or region of a clip without regenerating the whole thing — timestamp-level editing turns a full re-roll into a targeted fix, saving both time and credits on iterative work.
ByteDance names two specific upgrades over 2.0: the physical plausibility of complex motions, and the stability of scenes involving interactions among multiple subjects. Paired with director-style camera control, detailed shot direction carries through reliably.
Synchronized audio generates in the same pass as the video, with dialogue support across 10+ languages — built for advertising, e-commerce, and multilingual localization without a separate voice pipeline.
Seedance 2.5 fits wherever teams need longer, more consistent directed video without stitching multiple clips together.
Add a node, describe your scene, and generate up to 30 seconds in one pass.
Add a Seedance 2.5 node to the canvas and describe your scene in a text prompt, or upload a still to animate. Optionally add up to 50 reference items combined — 30 images, 10 video clips, 10 audio clips — to guide character, environment, or sound.
Set duration up to 30 seconds in a single continuous pass. Native audio generates alongside the video in 10+ languages. Director-style camera direction in the prompt carries through reliably.
Use timestamp-level regional editing to adjust one moment without regenerating the whole clip, chain into an upscaler or a voice node for extra narration, then export with commercial rights on your plan.
Real generated videos with the exact prompts — copy any prompt and run it yourself.
One subscription across 30+ AI models — no per-tool credit balances or separate signups. Credit cost per generation is shown live in the node before you run it.
500 weekly credits to test the model and see the output quality. No credit card required.
45,000 monthly credits with cheaper per-credit pricing, commercial rights, and video download.
90,000 monthly credits, up to 10 seats, shared workspace, centralized billing, and priority support.
210,000 monthly credits, API and integration support, unlimited seats, and dedicated management.
Full specs for the Seedance 2.5 model.
How Seedance 2.5 compares with other AI video models in the Phygital+ catalog.
Chain image, video, upscaling, and audio into one repeatable workflow.
Design the first frame with an image model, then animate it with Seedance 2.5 in one workflow.
Generate with Seedance 2.5, then upscale to delivery resolution for social, ads, or broadcast.
Add extra narration or a music track with a voice or audio node to finish the clip.
Write the script and shot list with a text model, then feed it straight into video generation.
Everything you need to know about Seedance 2.5 in Phygital+.
Seedance 2.5 is ByteDance's next-generation AI video model, officially released July 31, 2026 after a June 23 preview at the Volcano Engine FORCE conference. It generates up to 30 seconds of synchronized audio-video in a single continuous pass — double Seedance 2.0's native duration — with up to 50 multimodal reference inputs and timestamp-level regional editing. ByteDance skipped version numbers 2.1–2.4 to signal a generational jump.
Usage is credit-based and included in every Phygital+ plan — from the free plan up to Enterprise. The exact credit cost per generation is shown live in the node before you run it and varies with duration and resolution. ByteDance has not published official per-second API pricing for Seedance 2.5 as of its launch; this is separate from Phygital+'s own flat credit system.
Outputs created on any paid plan (Starter and up) come with commercial-use rights. As with Seedance 2.0, avoid generating recognizable real people without rights clearance — realistic depictions of public figures on this model line have previously raised copyright and likeness concerns. Check the model card for any provider-specific limits.
Seedance 2.5 is made by ByteDance's Seed research team, the same team behind Seedance 2.0. It launched first as the default video model inside ByteDance's own consumer apps, Jimeng (Dreamina) and Doubao Pro, with API access via BytePlus ModelArk and Volcano Engine opening in the first week of August 2026.
Up to 30 seconds in a single continuous pass — double Seedance 2.0's native 15-second ceiling — with multi-round extension available for anything longer. ByteDance names this "Long-Form Shot Consistency": steadier characters and more complex sequences held across the full 30 seconds without stitching separate generations together.
480p and 720p. This is an important correction to flag: ByteDance's June 23 preview announcement promised native 4K output, but what actually shipped on July 31 tops out at 720p — lower than Seedance 2.0's up-to-1080p ceiling. The resolution gain didn't materialize in exchange for the longer duration and larger reference budget; verify current specs before promising 4K to a client.
Yes — native synchronized audio generates in the same single pass as the video, with dialogue support across 10+ languages. ByteDance also highlights improved physical plausibility for complex motion and steadier scenes with multiple interacting subjects as named upgrades over 2.0.
Up to 50 in a single pass — 30 images, 10 video clips, and 10 audio clips combined — more than four times Seedance 2.0's 12-reference budget (9 images, 3 video, 3 audio). This supports richer character, brand, and scene consistency across a single long-form generation.
Yes. ByteDance opened API access via BytePlus ModelArk and Volcano Engine in the first week of August 2026, shortly after the July 31 consumer launch. In Phygital+, you can run Seedance 2.5 as a node on any plan, and API and integration support is available on the Enterprise plan.
Running Seedance 2.5 in Phygital+ means no separate account or credit balance for one model: it sits on the same canvas as 30+ other AI tools. Chain image generation, video, upscaling, and voice into one repeatable workflow under a single subscription, without waiting on ByteDance's own regional and channel rollout.
Browse the full catalog of 30+ AI models available in Phygital+.
Join 100+ teams using Phygital+ – every model in one workspace
Try Phygital+ free
