TokenRa
中文Get API Key

Video model · ByteDance

Seedance 2.5 Flash API: Pricing and How to Call It

Seedance 2.5 Flash is the speed-focused tier of ByteDance's Seedance 2.5 video model, called as an asynchronous task on TokenRa. Give it text or reference images and it returns a video with an audio track; resolution goes at the request top level, 480p or 720p, and billing is per video token.

480p
$6.522per million video tokens
720p
$6.522per million video tokens
Billing basis
Per tokenssettled against usage.total_tokens
Model ID
seedance-2.5-flash
Provider
ByteDance
Input
Text, reference images
Output
480p / 720p video
Mode
Async task
Audio
On by default

Overview

What it is

Flash is the tier of the Seedance 2.5 generation that leans on speed and cost. It shares one request format with 2.5 Pro: resolution, ratio, duration and generate_audio all go at the top level of the JSON, rather than inside metadata the way 2.0 does. That is the easiest place to trip between 2.5 and 2.0 — the two request bodies cannot be swapped.

Generation runs as an asynchronous task: POST to create the task and get a task_id, then poll the query endpoint and read the video from data.result_url once data.status is SUCCESS. For reference, a measured 2.5 Pro 4 to 5 second 720p video takes roughly 5 minutes.

  • To move the same code between the fast tier and the flagship tier, change only the model field. Within the 2.5 family, only Flash supports 480p — Seedance 2.5 Pro accepts 720p and 1080p only.
  • When you need reference images to pin a character or packaging look, pass image URLs in the images array.
  • When aspect ratio or duration matters, send ratio and duration explicitly instead of relying on defaults.

Pricing

Seedance 2.5 Flash pricing

Billing on TokenRa is per video token. Flash has the same unit price at 480p and 720p, settled against usage.total_tokens in the response. Token count rises with resolution and duration — for the same clip length, 720p carries noticeably more tokens than 480p, so the total cost per clip is still higher at the higher resolution.

ResolutionInput priceOutput priceNotes
480p$6.522 / million tokens$6.522 / million tokensFirst choice for everyday batch generation
720p$6.522 / million tokens$6.522 / million tokensFor deliverable output where quality matters

The live price in the Model Square is authoritative and this page is kept in step with it; if the two disagree, the Model Square wins — that would be a publishing error, so tell us and we will correct it.

Integration

How to call it

POSThttps://tokenra.io/v1/video/generations
Creates the task and returns task_id immediately; the video does not come back in this call.
GEThttps://tokenra.io/v1/video/generations/{task_id}
Queries the task; when data.status is SUCCESS the video link is in data.result_url.

Text to video

Output settings go directly at the top level:

POST /v1/video/generations
curl -X POST https://tokenra.io/v1/video/generations \
  -H "Authorization: Bearer $TOKENRA_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "seedance-2.5-flash",
    "prompt": "A paper boat drifting down a rainy neon street, cinematic close-up",
    "resolution": "720p",
    "ratio": "9:16",
    "duration": 4,
    "generate_audio": true
  }'

Reference image to video

Images in images are passed to the model as reference images, usually to pin a character, product or packaging look:

POST /v1/video/generations
curl -X POST https://tokenra.io/v1/video/generations \
  -H "Authorization: Bearer $TOKENRA_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "seedance-2.5-flash",
    "prompt": "The logo card slowly pushes in while a soft light sweeps across the frame",
    "images": ["https://example.com/logo-card.png"],
    "resolution": "720p"
  }'

Polling for the result

GET /v1/video/generations/{task_id}
curl https://tokenra.io/v1/video/generations/task_FUhA4FK80TV1p70mpHlyZXCxeHR0dPPa \
  -H "Authorization: Bearer $TOKENRA_API_KEY"

Poll every 10 to 15 seconds; do not turn it into a tight loop. data.status moves through QUEUED, IN_PROGRESS, SUCCESS or FAILURE; on failure the reason is in fail_reason and the pre-charged amount is refunded.

Request

Request parameters

ParameterTypeRequiredNotes
modelstringYesUse seedance-2.5-flash on this page.
promptstringYesScene, motion and camera description; may be omitted when content carries a text item.
resolutionstringYesTop-level field, 480p or 720p. Writing it into metadata is rejected.
ratiostringNoAspect ratio, such as 16:9 or 9:16.
durationintegerNoVideo length in seconds; a default applies when omitted.
generate_audiobooleanNoWhether to generate an audio track, on by default; pass false for a silent video.
imagesstring[]NoArray of reference image URLs.
contentarrayNoItem-by-item control of the material; can replace prompt and images.
What does not work: the 2.5 interface this page is based on reads output settings only from the top level; metadata is ignored, and sending resolution only inside metadata returns invalid_resolution. Video and audio references (video_url, audio_url) raise unsupported content type; fields such as seed, watermark and callback_url are dropped silently without an error, so do not count on them taking effect.

Errors

Common errors

HTTPErrorCause and handling
400invalid_resolutionNo resolution, a value outside 480p/720p, or it was written only into metadata. Move it to the top level and resend.
400prompt is requiredNo prompt, and no text item inside content.
401Invalid tokenAPI key missing, wrong or deleted; check the Authorization: Bearer header.
403insufficient_user_quotaBalance or plan quota is too low; top up and retry.
429—Rate limited; retry with exponential backoff.
503model_not_foundThe model ID does not exist or the current group has no usable channel; check with GET /v1/models.

Comparison

How it differs from 2.5 Pro and 2.0 Fast

2.5 Flash2.5 Pro2.0 Fast
Generation2.52.52.0
PositionFaster, cheaper tierFlagship qualityPrevious-generation fast tier
Where output settings goRequest top levelRequest top levelInside metadata
Resolution480p / 720p720p / 1080p2.0 tiers
AudioOn by defaultOn by default2.0 rules
Can request bodies be swappedYes, with ProYes, with FlashNot interchangeable with 2.5

FAQ

Frequently asked

What is Seedance 2.5 Flash?

It is the speed-focused tier of ByteDance's Seedance 2.5 video model, part of the 2.5 generation. It takes text and reference images, returns a video with an audio track, and is called as an asynchronous task on TokenRa.

How much does calling Seedance 2.5 Flash cost on TokenRa?

Billing is per video token, settled against usage.total_tokens, at $6.522 per million tokens for both 480p and 720p. The price on this page matches the live price in the Model Square.

How do I call Seedance 2.5 Flash?

Video generation is asynchronous. POST to create the task and get a task_id, then poll GET /v1/video/generations/{task_id} every 10 to 15 seconds until data.status is SUCCESS; the video link is in data.result_url.

Where do resolution and duration go?

At the top level of the request, the same as Seedance 2.5 Pro. The 2.5 generation does not read metadata; sending resolution only inside metadata is rejected with invalid_resolution.

How long does one video take?

A measured Seedance 2.5 Pro 4 to 5 second 720p video takes roughly 5 minutes; Flash is positioned as the faster tier, and actual time depends on queueing and resolution. Poll every 10 to 15 seconds.

How is Seedance 2.5 Flash different from 2.5 Pro and 2.0 Fast?

Flash and Pro are both in the 2.5 generation with the same request format and output settings at the top level; they differ in speed and price tier. 2.0 Fast is the previous generation and takes resolution, ratio and duration inside metadata, so the two request bodies cannot be mixed.

Is result_url a permanent link?

No. result_url is a temporary link; download and store the file yourself after the task succeeds.

"Seedance" and related names may be trademarks of their respective owners. This page is produced independently by TokenRa and offers model information and usage guidance only; it is not an official page of, or an authorization statement from, the rights holder.