Seedance 2.5 Flash Guide: API Access, Pricing and Limits

Last updated: November 15, 2026

Seedance 2.5 Flash is ByteDance’s faster, cheaper tier of the Seedance 2.5 video model. Announced November 12, 2026, it opened for API calls on BytePlus ModelArk and Volcano Ark on November 15. It generates clips of 4 to 30 seconds at 480p or 720p, with native audio, editing and extension.

This guide is based on the BytePlus Seedance 2.5 series tutorial, the Create video generation task API reference, the ModelArk pricing page, the Volcano Ark Seedance 2.5 tutorial and Volcano Engine’s launch announcement as reported by IT之家 on Sina Tech. For ByteDance’s image model, see our Seedream 5.5 guide.

Contents

What is new
Flash vs. Seedance 2.5 and 2.0
Pricing and access
How to make your first API call
Editing, extension and task rules
Limits to plan around
Frequently asked questions

What is new

Volcano Engine, ByteDance’s cloud business, introduced Seedance 2.5 Flash on November 12, 2026. According to the launch report, the company describes it as the model closest in quality to Seedance 2.5, with generation speed doubled and a lower price. It says Flash outperforms the previous flagship, Seedance 2.0, overall.

The announcement put the price for 720p text-to-video and image-to-video at about 0.99 yuan per second in China, roughly 34% below Seedance 2.5. It also said Flash would roll out to Jimeng, LibTV, Updream, Creao AI and TapNow, with the API arriving on Volcano Ark on November 15.

As of this update, both the BytePlus API reference and the Volcano Ark reference state that Seedance 2.5 and Seedance 2.5 Flash are fully available for API calls and online trials.

Flash vs. Seedance 2.5 and 2.0

The BytePlus capabilities overview lists these differences:

	Seedance 2.5 Flash	Seedance 2.5	Seedance 2.0
BytePlus model ID	`dreamina-seedance-2-5-flash-261115`	`dreamina-seedance-2-5-260628`	`dreamina-seedance-2-0-260128`
Output resolution	480p, 720p (8-bit)	480p, 720p (8-bit); 1080p (10-bit)	480p, 720p, 1080p; 4K (10-bit)
Output duration	4–30 seconds, or -1 (model chooses)	4–30 seconds, or -1	4–15 seconds, or -1
Reference assets per request	20 (10 images, 5 videos, 5 audio clips)	50 (30 images, 10 videos, 10 audio clips)	15 (9 images, 3 videos, 3 audio clips)
Audio-only reference	Yes	Yes	No (needs an image or video)
Draft mode	No	Yes	No
Output formats	mp4, mov	mp4, mov	mp4

All three support text-to-video, first-frame and first-last-frame generation, editing, extension, returning the last frame and generated audio. The 2.5 series adds native multilingual support for 11 languages, including English, Chinese, Spanish, Japanese and Korean. The new mov output uses H.264 video and PCM audio for better fidelity in editing work.

For a Google alternative, compare our Gemini Omni 1.2 Flash guide and the Gemini video generation limit answer.

Pricing and access

BytePlus prices video models in USD per million tokens. Per the pricing page, Flash costs USD 7.0 per million tokens for 480p and 720p online inference without video input, and USD 4.3 when the input includes video. The flagship costs USD 10.70 and USD 6.40 at the same resolutions. Offline (batch) inference is not supported for either model.

Estimated tokens are calculated as (input video duration + output video duration) × width × height × frame rate ÷ 1024. Actual usage is reported in usage.completion_tokens. BytePlus’s own examples for a 5-second 16:9 clip without video input:

Model	480p	720p	1080p
Seedance 2.5 Flash	USD 0.34 (0.07/second)	USD 0.76 (0.15/second)	Not supported
Seedance 2.5	USD 0.514 (0.103/second)	USD 1.156 (0.231/second)	USD 2.843 (0.569/second)

You are charged only for successfully generated videos. When the input includes video, a minimum token charge applies based on resolution, aspect ratio and duration.

Before you can enable any Seedance 2.5 series model, BytePlus requires an account balance above USD 30, an AI Savings Plan at the USD 30 tier or higher, or a Seedance 2.5 series resource pack. On Volcano Ark, the tutorial sets the equivalent threshold at a balance above 200 yuan, a 200-yuan savings plan or a resource pack.

How to make your first API call

Sign in and meet the activation rule: top up, buy a savings plan or buy a resource pack as described above.
Create an API key: on the ModelArk API Key management page, click Create API Key and store it as ARK_API_KEY.
Activate the model: enable Seedance 2.5 Flash in the console’s model activation page.
Submit a task: send a POST request to https://ark.ap-southeast.bytepluses.com/api/v3/contents/generations/tasks with "model": "dreamina-seedance-2-5-flash-261115", a content array with your text prompt, and options such as resolution, ratio, duration, generate_audio (default true) and watermark. In mainland China, use ark.cn-beijing.volces.com and doubao-seedance-2-5-flash-261115.
Poll or use a callback: the call returns a task ID. Retrieve the task until it succeeds, or set callback_url to receive status updates. Set return_last_frame to true if you plan to chain clips.
Retrieve results in time: standard video task IDs are valid for seven days after creation.

If you prefer not to code, the ModelArk console offers templates, and API Explorer has preset parameter templates for one-click test calls.

Editing, extension and task rules

The 2.5 series infers the task type from your inputs and prompt, and the tutorial warns that breaking task constraints causes errors:

Editing: include at least one reference_video of 4 to 30 seconds, set ratio to adaptive and duration to -1, and use words such as add, remove or replace. Output keeps the source video’s aspect ratio and length.
Extension: include a reference video, set ratio to adaptive and ask to extend forward or backward or continue the story.
First frame or first and last frame: ratio must be adaptive.
Catch errors early: set omni_reference_task_type to edit or extend so the API validates constraints at submission. Mismatches still return InvalidParameter.TaskTypeMismatch.

Limits to plan around

No 1080p or 4K. Use the flagship for 1080p delivery.
No Draft mode. You cannot preview a 480p draft and then render the same result at higher resolution.
Input caps. Reference videos total 30 seconds or less and each must be 200 MB or less; audio clips must be wav or mp3 files of 15 MB or less.
Task timeout. Tasks expire after 48 hours by default, adjustable from 1 to 72 hours with execution_expires_after.
Vendor claims. The doubled speed and 34% price cut come from Volcano Engine and have not been independently verified.

If you only need occasional clips in a chat app, our answers on whether ChatGPT can generate videos cover consumer options.

Frequently asked questions

What is Seedance 2.5 Flash?

Seedance 2.5 Flash is a faster, lower-cost tier of ByteDance’s Seedance 2.5 video generation model. Volcano Engine announced it on November 12, 2026, and says it is the model closest in quality to Seedance 2.5, with generation speed doubled.

When did the Seedance 2.5 Flash API open?

The announcement said the API would go live on Volcano Ark on Sunday, November 15, 2026. Both the BytePlus ModelArk and Volcano Ark API references now state that Seedance 2.5 and Seedance 2.5 Flash are fully available for API calls and online trials.

What is the Seedance 2.5 Flash model ID?

On BytePlus ModelArk the model ID is dreamina-seedance-2-5-flash-261115. On Volcano Ark in mainland China it is doubao-seedance-2-5-flash-261115. Both use the same Create video generation task API.

How much does Seedance 2.5 Flash cost?

BytePlus lists USD 7.0 per million tokens for 480p and 720p output without video input, and USD 4.3 per million tokens when the input includes video. Its billing example puts a 5-second 16:9 clip at about USD 0.34 at 480p and USD 0.76 at 720p.

Does Seedance 2.5 Flash support 1080p or Draft mode?

No. Flash outputs only 480p and 720p, both at 8-bit color depth. Draft mode, which previews a 480p version before a final render, is supported only by the flagship Seedance 2.5.

Seedance 2.5 Flash: Power-User Guide to the API, Pricing and Limits