Seedance Version History: 1.0 to 2.5 Timeline (2026)

Seedance 2.5 is the latest version of Seedance verified as of October 21, 2026. ByteDance officially launched it on July 31, 2026. Seedance version history begins in June 2025 with Seedance 1.0, followed by 1.5 Pro, 2.0, and 2.5. For developers, these generation names must be matched to the exact model IDs in the API catalog.

Last verified: 2026-10-21

Seedance Version History: Timeline and API Models

The timeline below separates model generations from API variants. A generation describes a milestone in the Seedance family; an API model ID identifies the model selected in a request. Multiple IDs within one generation do not establish separate release dates.

Version or variant	Release or announcement date	Key change or distinction	API model ID
Seedance 1.0	2025-06	Text- and image-driven video generation, including multi-shot storytelling.	No entry in the reviewed catalog
Seedance 1.5 Pro	2025-12-16	Joint audio-video generation, with synchronized speech and expanded camera control.	`seedance-1.5-pro`
Seedance 2.0	2026-02-12, official launch announcement	A unified architecture supporting text, image, audio, and video inputs.	Select a variant below
Seedance 2.0 Pro	Not officially dated	A separately selectable API variant within the 2.0 generation.	`seedance-2.0-pro`
Seedance 2.0 Fast	Not officially dated	A separately selectable API variant within the 2.0 generation.	`seedance-2.0-fast`
Seedance 2.0 Mini	Not officially dated	A separately selectable API variant within the 2.0 generation.	`seedance-2.0-mini`
Seedance 2.5	2026-07-31	Longer storytelling, expanded references, and more precise editing.	`seedance-2.5`

The June 2025 starting point is consistent with ByteDance’s June 11 technical-report announcement. The 1.5 Pro post explicitly announces its launch on December 16. For 2.0, February 12 is the date of the official launch post, which describes the model as recently launched. Separate dates for the three 2.0 API variants were not established in the reviewed official sources.

Model IDs are reproduced from the API models and pricing reference. They identify this API’s catalog entries and should not be treated as universal identifiers across providers.

What Changed Across Seedance Generations?

Seedance 1.0: The Video Generation Foundation

Seedance 1.0 established the family’s foundation in text-to-video and image-to-video generation. ByteDance’s technical announcement describes 1080p output and transitions between multiple shots. That makes 1.0 an important historical reference when reading early examples or evaluating how the family developed.

For a creator reviewing an older tutorial, the practical question is which task it demonstrates. A tutorial about composing shots can remain useful even when its model selection instructions need updating. Keep the creative lesson separate from the integration details: a prompt structure may transfer, while a model identifier must be checked against current documentation.

Seedance 1.5 Pro: Sound Joins the Picture

Seedance 1.5 Pro added joint audio-visual generation. ByteDance’s launch post emphasizes synchronization between sound and visuals, multilingual speech, and more expressive camera movement. This is a meaningful milestone for understanding why later Seedance workflows discuss dialogue, performance, and background sound together.

When comparing examples from this generation, listen as well as watch. Consider whether spoken words fit the intended scene, whether timing supports the action, and whether the camera movement serves the story. For a focused comparison with the current generation, see Seedance 1.5 Pro vs 2.5.

Seedance 2.0: A Broader Reference Workflow

Seedance 2.0 brought text, images, audio, and video into a unified multimodal generation architecture. Its official presentation emphasizes using references to guide performance, composition, and camera movement. The historical change is therefore broader than a higher version number: more kinds of creative material can inform the result.

For example, a creative brief might use an image to define a product and a video to communicate a camera move. Those materials serve different purposes, so describe each role clearly. This is a planning example, not a measured generation result. The Seedance 2.0 model overview provides further background.

Seedance 2.5: Longer Stories and More Targeted Control

ByteDance’s Seedance 2.5 release announcement describes generation of up to 30 seconds in a single pass, expanded multimodal referencing, and timestamp-based editing. Its project page also emphasizes longer narratives and more precise reference interpretation. These are model-level capabilities; the controls exposed by a particular integration still need checking.

For planning, divide a scene into a beginning, a visible change, and an ending. A product demonstration, for example, might introduce an object, show one action, and finish on a clear result. Longer duration is useful when those beats need room, but a simple idea can still work better as a short clip.

Seedance 2.0 vs 2.5 API Models: Read the Names Correctly

The most useful distinction for developers is between a family label and a request value. “Seedance 2.0” identifies a generation, while its Pro, Fast, and Mini entries are distinct catalog choices. Seedance 2.5 has its own entry. Copy the exact identifier rather than deriving one from a heading or changing punctuation by hand.

Suffixes alone do not establish measured speed, output quality, or cost. A sensible comparison starts with your actual workload: the same subject, intended duration, aspect ratio, and evaluation criteria. Then compare results under supported settings. Avoid calling a model the best choice for every task based only on its position in the timeline.

Use the Seedance 2.0 API guide for integration context. This history explains how the generations relate; the current API documentation determines how to construct a valid request.

Resolution Is a Separate Selection Decision

The current Seedance 2.5 creator lists 480p, 720p, and 1080p output. Choose among those settings according to the intended delivery format. For API requests, retrieve the selected model’s supported settings before submission rather than assuming that website controls map directly to every endpoint.

A version number is not a resolution specification. When reviewing an exported clip, record the chosen setting and inspect the downloaded file. This makes a comparison reproducible and helps distinguish creative improvements from differences caused by output size, compression, or the display used for review.

How to Use This Timeline When Updating a Workflow

A version history becomes useful when it helps you decide what to preserve and what to retest. Start with one representative task from your existing workflow, such as a product close-up or a dialogue scene. Write down the qualities that matter before looking at the newer output.

Record the baseline: Save the prompt, reference assets, model identifier, and selected settings together.
Check compatibility: Review supported modes and parameters for the target model before adapting the request.
Keep the brief consistent: Use the same creative objective so that a different prompt does not obscure the model comparison.
Inspect the whole clip: Review subject consistency, action, framing, and sound wherever sound is part of the brief.
Document the decision: Explain which observed result justifies keeping or changing the model.

These steps are an evaluation method, not a claim that every newer version improves every scene. A useful migration record tells another person why the selected model fits the task. It also makes later troubleshooting easier because the inputs and decision criteria remain visible.

For implementation, begin with the API documentation. If authentication is your starting point, follow the Seedance API key guide. Keep the historical timeline alongside your integration notes, while treating current capability information as the reference for supported requests.

Frequently Asked Questions

What is the latest version of Seedance?

Seedance 2.5 is the latest version verified on October 21, 2026. ByteDance’s current model catalog features it, and its official launch announcement is dated July 31, 2026. The verification date matters because a version-history article is a snapshot of the evidence checked.

Is Seedance 2.5 out yet?

Yes. ByteDance officially launched Seedance 2.5 on July 31, 2026. The reviewed API catalog also includes its model identifier. An official model launch and an individual provider’s supported controls are separate facts, so consult the integration documentation for the workflow you intend to use.

When was Seedance first released?

Seedance was first released in June 2025. ByteDance’s June 11, 2025 technical-report announcement documents Seedance 1.0 and its text- and image-driven video capabilities. The timeline uses the month for the first release rather than equating a technical-report publication date with an exact initial launch day.

Are Seedance 2.0 Pro, Fast, and Mini separate generations?

They are separately named API variants within the Seedance 2.0 generation. Their catalog entries do not establish a chronological sequence between them. Separate official launch dates were not verified, so the timeline marks each variant “Not officially dated.”

Does upgrading to Seedance 2.5 only require changing the model ID?

Review the request as well as the identifier. Modes, inputs, and output settings must match the target model’s supported capabilities. Keep a representative baseline example, test the revised request, and inspect the resulting clip before applying the change across an established workflow.

Try Seedance 2.5 Today

Open the Seedance AI video generator and start with a clear scene you can evaluate. Choose Seedance 2.5, define the action and framing, and select a supported output resolution. Developers can use the timeline above to identify the generation, then consult the API reference for the exact request configuration.