# Seedance 2.5 Flash pricing: official rates and clip costs

By SeedRouter · Published 2026-11-12 · Updated 2026-11-12

**Seedance 2.5 Flash costs $7.00 per million video tokens without video input, or $4.30 per million video tokens with video input, in BytePlus's official price table checked on November 12, 2026.** Both rates cover 480p and 720p. They are official reference prices, not SeedRouter selling prices and not fixed prices per video. \[1]

The lower rate for video input is easy to misread. A reference video adds billable duration, and a minimum token charge can apply. As a result, attaching a reference can increase the total bill even though the rate per token falls.

This guide turns the published rates into explicit examples, separates estimates from actual charges, and shows how to compare the same job across models. SeedRouter pricing for Flash is pending. For access timing, see the [Flash release-date guide](https://seedrouter.ai/blog/seedance-2-5-flash-release-date); for capabilities and current status, see the [model page](https://seedrouter.ai/models/seedance-2-5-flash).

## Official Seedance 2.5 Flash pricing

BytePlus publishes these online-generation rates in USD per million video tokens. The reference-video column applies to requests that include video input. \[1]

| Model              | Without video input | With video input | Resolution covered |
| ------------------ | ------------------- | ---------------- | ------------------ |
| Seedance 2.5 Flash | $7.00               | $4.30            | 480p and 720p      |
| Seedance 2.5       | $10.70              | $6.40            | 480p and 720p      |

The table is a dated comparison of official BytePlus rates. It does not quote a SeedRouter discount, a subscription allowance or another application's credit price. Keep those categories separate when budgeting: the same model name can appear in products with different commercial terms.

Also keep the billing unit attached to every number. “Seven dollars” is incomplete; “seven dollars per million video tokens without video input” tells you how to use the figure. The resolution labels describe supported output classes. They do not mean a 480p clip and a 720p clip consume the same number of tokens.

## How do video tokens become a clip price?

BytePlus's estimation formula uses the output dimensions, frame rate and combined input-video and output-video duration. The estimated price is the token count multiplied by the applicable rate, with the million-token unit accounted for. \[1]

```text
estimated video tokens =
  (input video seconds + output video seconds)
  × output width × output height × output frames per second / 1024

estimated USD cost = estimated video tokens × USD rate / 1,000,000
```

For a request without video input, the input-video term is zero. For a request with video input, use the applicable combined video duration and check the minimum-token rules before treating the formula as the bill. The official documentation identifies the returned token usage as the basis for actual consumption. \[1]

The practical consequence is that a clip's duration alone is insufficient for a precise estimate. “A 10-second video” leaves out resolution, dimensions and input mode. A useful budget line records those settings together, so another person can reproduce the estimate rather than guess what it includes.

An image reference is also different from a reference video for choosing these two rate columns. The heading is about video input. Do not select the video-input rate merely because a request contains some kind of reference material.

## Worked 720p examples without video input

For the following arithmetic, assume **1280 × 720 pixels, 24 frames per second and no video input**. These are explicit example settings, not a claim that every aspect ratio produces those dimensions or that every service uses them as defaults.

At those settings, one second represents 21,600 estimated video tokens. A five-second clip represents 108,000 tokens, which at the Flash rate gives an unrounded estimate of $0.756. BytePlus's own five-second, 16:9, 720p example rounds to $0.76. \[1]

| Output duration | Estimated video tokens | Calculated Flash cost, rounded to cents |
| --------------- | ---------------------- | --------------------------------------- |
| 5 seconds       | 108,000                | $0.76                                   |
| 10 seconds      | 216,000                | $1.51                                   |
| 30 seconds      | 648,000                | $4.54                                   |

The 10-second and 30-second rows are arithmetic estimates under the same assumptions. They are not invoices from generated videos. A different pixel size, frame rate or input mode changes the calculation. Keep full precision while estimating a batch and round the final total, rather than repeatedly multiplying a rounded per-second price.

For instance, 100 ten-second clips under these assumptions total an estimated $151.20 before any applicable commercial adjustments. Multiplying the displayed $1.51 by 100 instead produces $151.00. The difference is only rounding, but it illustrates why the underlying token calculation is the better budgeting unit.

At 480p, BytePlus's official five-second, 16:9 example is approximately $0.34. \[1] Use that as a separate documented example rather than assuming every output carrying a “480p” label has an identical price. The general formula still depends on actual dimensions.

## Why can a reference video make the bill larger?

Video input changes both sides of the calculation: the token rate falls, while the billable duration grows. Comparing only $4.30 with $7.00 misses the second change.

Consider a ten-second output at the 720p settings above. Without video input, the estimated charge is $1.512 before rounding. Add a ten-second reference video and the combined duration becomes 20 seconds. That produces 432,000 estimated tokens at $4.30 per million, or **$1.8576 before any applicable minimum-token adjustment**.

The reference-video estimate is already larger than the no-video estimate. This arithmetic does not imply that a reference is wasteful: it may be essential to the requested result. It means the reference should be chosen for the control it provides, with its cost included from the start.

Minimum billing is another reason not to assume a tiny reference gives a discount. BytePlus states that video-input requests in the Seedance 2.5 series have a minimum token consumption, determined by resolution, aspect ratio and output duration. If the estimate falls below that threshold, the minimum applies. \[1]

Check the linked minimum-token table for the exact configuration instead of extrapolating a universal threshold from one example. This guide does not claim that a short reference reduces the bill, nor that the raw formula alone predicts every video-input charge.

## How much cheaper is Flash than the flagship?

Compared with [Seedance 2.5](https://seedrouter.ai/models/seedance-2-5), for the same number of billable tokens at 480p or 720p, the official no-video rate falls from $10.70 to $7.00: approximately **34.6% lower**. The video-input rate falls from $6.40 to $4.30: approximately **32.8% lower**. Those percentages are calculations from the published rate table. \[1]

They compare matched token counts, not every possible finished-video workflow. Resolution, reference duration, minimum billing and the number of attempts can all affect a project's total. A low rate is useful only if the model can deliver an acceptable result within the settings you need.

Flash's published output options are 480p and 720p. \[2] If your deliverable needs a higher native resolution, do not compare a Flash estimate directly with a higher-resolution flagship quote and describe the difference as a like-for-like saving. First decide the required output, then compare options that satisfy it.

For an actual evaluation, record both the total spend and the number of usable clips. Dividing one by the other gives a useful cost-per-accepted-result measure. We have not generated a benchmark set for Flash, so there is no measured acceptance rate or production-cost advantage reported here.

## A practical budget checklist

Before approving a batch, record the model, service, currency, output dimensions, frame rate, duration and whether video is supplied as input. Add the applicable minimum-token rule. This is enough information to explain why one estimate differs from another.

Keep the planned attempts separate from the deliverable count. If you need ten finished clips but expect to try several prompts for each, a budget for ten generations may not describe the work you intend to do. Use your own observed acceptance rate when available; do not borrow an unsupported success percentage from a model announcement.

Finally, compare the estimate with the service's charge information before submitting. On SeedRouter, Flash selling prices have not been published yet. The [Flash model page](https://seedrouter.ai/models/seedance-2-5-flash) labels the BytePlus figures as official reference pricing so they cannot be mistaken for a current SeedRouter quote.

## Frequently asked questions

### Are $7.00 and $4.30 official prices?

Yes. They are BytePlus's published USD rates per million video tokens, checked November 12, 2026. They apply to different input modes at 480p and 720p. They are not prices for a single clip and are not SeedRouter selling prices. \[1]

### Does 720p cost the same as 480p?

The rate per million tokens is the same for those two output classes, but token consumption depends on dimensions. A larger output can therefore cost more despite sharing the same token rate. \[1]

### Is the reference-video rate always cheaper overall?

No. Input video adds billable duration, and minimum-token billing can apply. Compare the complete calculation, not just the two rate numbers. The worked example above shows a larger bill despite the lower per-token rate.

### Can I use these figures as my SeedRouter budget?

Use them to understand the official billing model. SeedRouter Flash pricing is pending, so they are not a quote for generation on SeedRouter. Check the model page for the current availability and pricing status.

## Plan from the complete calculation

Seedance 2.5 Flash pricing becomes easier to compare once the rate, dimensions, duration and input mode are written together. Use the official table as a dated reference, calculate without premature rounding, and account for video-input minimums. For the latest confirmed product information, return to the [Seedance 2.5 Flash page](https://seedrouter.ai/models/seedance-2-5-flash).

## Sources

1. [BytePlus: Model pricing — video rates, billing formula, minimum consumption and examples](https://docs.byteplus.com/en/docs/modelark/model-pricing), checked November 12, 2026.
2. [Volcano Engine: Doubao Seedance 2.5 Flash model details](https://ark.volcengine.com/region:cn-beijing/model/detail?name=doubao-seedance-2-5-flash), checked November 12, 2026.