Skip to main content

Seedance 2.5 API Pricing: 30-Second Cost Calculator

5 min readAI Video Generation

A 30-second Seedance 2.5 clip is not a flat-price product. Resolution, video input, billed successful attempts, and acceptance rate all change the real cost.

Seedance 2.5 API 30-second calculator connecting token rates, attempts, and accepted clips

A 30-second Seedance 2.5 video costs about $3.09 at 480p or $6.93 at 720p under one specific BytePlus scenario: 16:9, 24 fps, no video input, and one successful generation. Those numbers are useful baselines, not universal clip prices.

Write your request in one line before budgeting:

model ID + resolution + aspect ratio + output duration + video-input condition + expected successful attempts

If one field is missing, a headline “30-second cost” is too weak for a customer quote or annual plan.

Seedance 2.5 30-second cost depends on the billing contract, request settings, and successful attempts

Start with the billing owner, not the lowest number

The current BytePlus ModelArk pricing page lists dreamina-seedance-2-5-260628 for 480p and 720p online inference:

Request conditionOnline rateListed output
Input does not include video$10.70 / 1M tokens480p, 720p
Input includes video$6.40 / 1M tokens480p, 720p

These are token rates, not per-video prices. The video-input rate is lower, but its token calculation adds input-video duration and can apply a minimum-token floor. A lower rate therefore does not guarantee a cheaper request.

China ModelArk publishes a separate CNY contract and a different model label, doubao-seedance-2.5. Do not convert its CNY row with a live exchange rate and call it BytePlus pricing. Do not mix either direct API contract with a provider subscription, consumer credits, or a promotional balance.

The current direct price row lists only 480p and 720p for Seedance 2.5. If another platform advertises 1080p or 4K, verify whether it is offering a different model, an upscale step, or a provider-specific pipeline before comparing prices.

Calculate 30 seconds from the official 5-second example

BytePlus provides a no-video-input example fixed to 5 seconds, 16:9, and 24 fps:

ResolutionOfficial 5-second estimatePer-second reference30 seconds at the same settings
480p$0.514about $0.103about $3.09
720p$1.156about $0.231about $6.93

The reproducible shortcut is:

one 30-second attempt = official 5-second example × 6

Using the exact 5-second values gives $3.084 for 480p and $6.936 for 720p. The rounded per-second column gives $3.09 and $6.93. Use the exact row for spreadsheets and the rounded values for a quick budget.

The linked official Seedance 2.5 pricing calculator exposes output durations from 4 to 30 seconds, 480p/720p, multiple aspect ratios, and 24 fps. Use it when the request is not 16:9.

Official 5-second Seedance 2.5 examples expanded into 30-second 480p and 720p estimates

Use tokens for requests that do not match the table

ModelArk describes the estimate as:

video price ≈ token unit rate × token consumption

estimated tokens = (input-video duration + output duration) × output width × output height × frame rate ÷ 1024

With no video input, the first duration is zero. With video input, add the reference-video duration. Seedance 2.5 can also apply a minimum-token rule tied to resolution, aspect ratio, and output duration. If the formula returns less than that floor, the floor controls the estimate.

That is why comparing $10.70/M with $6.40/M in isolation is misleading. Put both configurations into the official calculator or run a small task and read usage.completion_tokens.

Once actual usage is available, reconcile it with:

request cost = completion_tokens ÷ 1,000,000 × the matching account rate

For example, 650,000 completion tokens would be about $6.96 at $10.70/M. The same token count would be $4.16 at $6.40/M, but video-input and no-video requests usually do not consume the same number of tokens.

Convert attempt cost into accepted-clip cost

The current ModelArk price page states that only successfully generated videos are charged; generation failures such as moderation failures are not. That is a valuable billing boundary, but a successful generation is not necessarily an accepted deliverable.

A technically successful clip can still fail character consistency, action, lip sync, typography, or brand requirements. It remains a billed success. Production budgeting needs:

cost per accepted clip = attempt estimate × billed successful attempts ÷ accepted clips

For a 30-second, 16:9, 720p request without video input:

Successful attempts needed per accepted clipEstimated cost
1$6.94
2$13.87
3$20.81

Suppose ten 720p attempts generate successfully and editors accept six:

  • estimated spend: 10 × $6.936 = $69.36;
  • accepted clips: 6;
  • estimated cost per accepted clip: $69.36 ÷ 6 = $11.56.

This is not a recommendation to generate a fixed number of times. Replace the example acceptance rate with your own history. If the workflow is new, run a small representative set, record successful attempts and accepted outputs, then update the budget.

Verify four things before integration

Model and account

Confirm the exact model ID, activation status, region, currency, and quota in the account you will bill. A public price row does not prove that every account already has access.

Request contract

Record resolution, aspect ratio, output duration, total reference-video duration, and frame rate. An official calculator offering 30 seconds does not prove that every provider, old SDK, or gateway has synchronized the same parameter range.

Failure contract

Verify how the route treats moderation, task failure, expiration, cancellation, and retries. The uncharged-failure rule above belongs to the current direct ModelArk contract. A third-party provider can have a different policy.

Usage evidence

Store the task ID, model ID, settings, success state, and completion tokens. Reconcile those values with the bill before treating an estimate as your internal unit cost.

For host, authentication, asynchronous task creation, polling, callbacks, and downloads, continue to the Seedance API guide. That page owns implementation; this pricing guide does not invent a Seedance 2.5 endpoint from older 2.0 examples.

The practical 30-second answer

Under the current BytePlus example for 16:9, 24 fps, and no video input:

  • 480p: about $3.09 per successful 30-second attempt;
  • 720p: about $6.93 per successful 30-second attempt.

Multiply by the average number of billed successful attempts required for one accepted clip. Use the official calculator for video input or another aspect ratio, and replace estimates with actual usage before committing a production or customer budget.

#Seedance 2.5#Seedance API Pricing#Video Cost Calculator#BytePlus ModelArk#AI Video Budget
Share: