# Cheapest Seedance 2.5 API: every platform priced per second

By TokLab · Published 2026-11-04 · Updated 2026-11-04

The lowest list price for the [Seedance 2.5](https://toklab.ai/docs/seedance-2-5) API is reapi.ai's Eco tier: $0.201 per second of 720p video and $0.347 at 1080p, an economy route that offers only those two resolutions. At ByteDance's own rate of about $0.231 a second at 720p, five platforms sell the standard model: BytePlus ModelArk (ByteDance's platform), OpenRouter, Replicate, Runware and TokLab. The most expensive platform we checked, fal.ai, charges $0.4730 for the same second.

Every price here comes from each platform's own pricing or model page on November 4, 2026. TokLab is our product. Among the platforms at ByteDance's rate it is the lowest by 0.2%, which is a rounding difference, not a reason to switch on its own.

<video src="/media/blog/2026-11-04-cheapest-seedance-2-5-api/seedance-2-5-720p.mp4" poster="/media/blog/2026-11-04-cheapest-seedance-2-5-api/poster.webp" aria-label="A 5-second 720p Seedance 2.5 clip rendered on TokLab, labelled with what it was billed and what the same clip costs at fal.ai's rate" preload="metadata" />

## Seedance 2.5 price per second on 10 platforms

Text or image to video, 16:9, no reference video, list price:

| Platform                                                                        |    480p |       720p |      1080p |    4K |
| ------------------------------------------------------------------------------- | ------: | ---------: | ---------: | ----: |
| [TokLab](https://toklab.ai/docs/seedance-2-5)                                                    | $0.1026 |    $0.2307 |    $0.5672 |     — |
| [BytePlus ModelArk](https://docs.byteplus.com/en/docs/ModelArk/1544106)         | $0.1028 |    $0.2312 |    $0.5686 |     — |
| [OpenRouter](https://openrouter.ai/bytedance/seedance-2.5)                      | $0.1028 |    $0.2311 |          — | $2.08 |
| [Replicate](https://replicate.com/bytedance/seedance-2.5)                       | $0.1028 |    $0.2312 |          — |     — |
| [Runware](https://runware.ai/seedance-2-5)                                      | $0.1028 |    $0.2311 |    $0.6085 |     — |
| [reapi.ai](https://reapi.ai/models/seedance-2-5) Eco                            |       — |     $0.201 |     $0.347 |     — |
| [reapi.ai](https://reapi.ai/models/seedance-2-5)                                |  $0.119 |     $0.267 |     $0.643 |     — |
| [Atlas Cloud](https://www.atlascloud.ai/models/seedance-2.5)                    | $0.1397 | not public | not public |     — |
| [kie.ai](https://kie.ai/seedance-2-5)                                           |  $0.140 |     $0.315 |     $0.790 |     — |
| [WaveSpeedAI](https://wavespeed.ai/models/bytedance/seedance-2.5/text-to-video) |   $0.18 |      $0.36 |      $0.90 | $1.80 |
| [fal.ai](https://fal.ai/models/bytedance/seedance-2.5/text-to-video)            | $0.2205 |    $0.4730 |     $1.164 |     — |

A dash means the platform does not list that resolution for Seedance 2.5. Notes on how each figure was read:

* **BytePlus** prices tokens: $10.70 per million at 480p and 720p, $11.70 at 1080p. The figures above are its own worked examples (5 seconds, 16:9) divided by five.
* **OpenRouter** routes to ByteDance's Seed service at the same $10.70 per million and quotes "$0.1028/second at 480p to $2.08/second at 4K". Its page does not break out 1080p.
* **TokLab** bills tokens at $10.681 and $11.671 per million. The per-second figures use ByteDance's frame sizes (854×480, 1280×720, 1920×1080) at 24 frames a second. TokLab stops at 1080p.
* **Runware** prints $0.11, $0.24 and $0.62 a second and says they "convert from per-token billing": $10.7 per million tokens at 480p and 720p, which is ByteDance's own rate, and $12.52 at 1080p. The table converts those token rates the same way as the other rows.
* **reapi.ai** lists two tiers: the standard model, which allows real-person references, and Eco, an economy tier at 720p and 1080p only. Credits are $0.001 each and failed generations are refunded.
* **Atlas Cloud** advertises "from $0.134/sec," 20% off $0.167. Its price calculator gives $0.1397 at 480p; we could not read 720p and 1080p from its public pages.
* **kie.ai** prices credits at $0.005: 28, 63 and 158 credits a second.
* **fal.ai** bills $0.0214 per 1,000 video tokens ($0.0234 at 1080p) and quotes these per-second figures itself. They run about 2% above its token rate at ByteDance's frame sizes (1280×720 works out to $0.4622), so the tables use fal.ai's own quotes.

Together AI lists Seedance 2.5 at "$0.115 per video" with no resolution or length, so it is left out.

![Seedance 2.5 price per second at 720p across API platforms, from reapi.ai Eco at $0.201 and TokLab at $0.2307 to fal.ai at $0.473](https://toklab.ai/media/blog/2026-11-04-cheapest-seedance-2-5-api/price-720p.webp)

## What a clip costs at each length

Seedance 2.5 renders anything from 4 to 30 seconds. At 720p, 16:9, without a reference video:

| Platform                                 |   5 s |  10 s |  15 s |        30 s |
| ---------------------------------------- | ----: | ----: | ----: | ----------: |
| TokLab                                   | $1.15 | $2.31 | $3.46 |       $6.92 |
| BytePlus, OpenRouter, Replicate, Runware | $1.16 | $2.31 | $3.47 | $6.93–$6.94 |
| reapi.ai Eco                             | $1.01 | $2.01 | $3.02 |       $6.03 |
| reapi.ai                                 | $1.34 | $2.67 | $4.01 |       $8.01 |
| kie.ai                                   | $1.58 | $3.15 | $4.73 |       $9.45 |
| WaveSpeedAI                              | $1.80 | $3.60 | $5.40 |      $10.80 |
| fal.ai                                   | $2.37 | $4.73 | $7.10 |      $14.19 |

A 30-second take on fal.ai costs about as much as two on BytePlus or TokLab.

## Prices with a reference video

Attach a reference video and every platform changes the rate, and most also bill the reference's length on top of the output. Per billed second:

| Platform     |    480p |    720p |   1080p | What is billed                 |
| ------------ | ------: | ------: | ------: | ------------------------------ |
| TokLab       | $0.0613 | $0.1378 | $0.3400 | input + output tokens          |
| Runware      | $0.0658 | $0.1479 | $0.3640 | input + output tokens          |
| reapi.ai Eco |       — |  $0.121 |  $0.207 | uploaded video at a lower rate |
| reapi.ai     |  $0.072 |  $0.161 |  $0.384 | uploaded video at a lower rate |
| kie.ai       |  $0.085 |  $0.190 |  $0.475 | input + output seconds         |
| WaveSpeedAI  |   $0.11 |   $0.22 |   $0.55 | input + output seconds         |
| fal.ai       | $0.1323 | $0.2838 | $0.6984 | input + output tokens          |

TokLab's rate is $6.38 per million tokens ($6.996 at 1080p) and Runware's $6.848 ($7.49); both are converted here the same way as above. OpenRouter starts at $0.06149 a second at 480p. fal.ai applies 0.6 times its normal price. Replicate lists $0.4304 at 480p and $0.9676 at 720p for its video-input variant, several times its plain price, so check its page before sending reference videos there.

A rate per billed second is not the price of the clip: a 5-second output with a 10-second reference bills 15 seconds on all of these.

## How Seedance 2.5 billing works

ByteDance prices Seedance 2.5 in video tokens. Its formula is (input video seconds + output seconds) × width × height × 24 ÷ 1,024. A second of 1280×720 is 21,600 tokens, so at $10.70 per million it costs $0.2311.

Platforms that print a price per second have done that conversion for you; BytePlus, OpenRouter, Runware, TokLab and fal.ai bill the tokens the render reports. The same request therefore costs the same everywhere a platform charges ByteDance's rate, and 4:3 or 1:1 clips cost a little less than 16:9 at the same resolution because they have fewer pixels. What is in the frame does not change the price.

BytePlus states that "you are only charged for successfully generated videos" and charges nothing when a generation fails, for example in content moderation. fal.ai and TokLab also skip failed generations. On other platforms, check the refund policy before you run volume.

## Two real bills

On November 4, 2026 we rendered one prompt with Seedance 2.5 on TokLab at 720p and at 480p, both 16:9 and 5 seconds (the clip above is the 720p one).

| Setting         | Tokens billed |    Charge |
| --------------- | ------------: | --------: |
| 720p, 16:9, 5 s |       108,900 |  $1.16316 |
| 480p, 16:9, 5 s |        48,437 | $0.517356 |

The 720p clip came back as 1280×720 with 121 frames, and 1280 × 720 × 121 ÷ 1,024 is exactly 108,900. Five seconds rendered one frame more than 5 × 24, which is 0.8% more tokens than ByteDance's 5-second example; with TokLab's slightly lower rate the bill lands 0.6% above that example's $1.156. The 480p clip shows a similar gap, 0.7%, against its $0.514. We read the 480p charge from the account balance; the 720p charge is its token count times $10.681 per million, rounded to the nearest $0.000002 as the ledger rounds every charge.

The same 720p clip at fal.ai's quoted $0.4730 a second would cost $2.37.

## Is Seedance 2.0 cheaper?

Yes, by about a third. On TokLab at 720p, per second without a reference video:

| Model             | 720p per second | 5-second clip |
| ----------------- | --------------: | ------------: |
| Seedance 2.5      |         $0.2307 |         $1.15 |
| Seedance 2.0      |         $0.1511 |         $0.76 |
| Seedance 2.0 Fast |         $0.1207 |         $0.60 |
| Seedance 2.0 Mini |         $0.0805 |         $0.40 |

Seedance 2.5 costs 53% more than 2.0 per second. What it adds is length (up to 30 seconds against 15) and larger reference sets. If your clips are 15 seconds or shorter and 2.0 looks good enough, it is the cheaper model. See [Seedance 2.0](https://toklab.ai/docs/seedance-2-0) for its rates on every resolution.

## How to pay less for Seedance 2.5

1. **Draft at 480p first.** Seedance 2.5 has a draft mode that renders a 480p preview at the 480p price; you finish only the takes you keep at 1080p. BytePlus, TokLab, Atlas Cloud and fal.ai expose it. A 1080p second costs 5.5 times a 480p one, so drafting pays off the first time it saves you a discarded 1080p render.
2. **Ship at the resolution you need.** Going from 720p to 1080p multiplies the price by about 2.5.
3. **Keep takes short while iterating.** Price is linear in length; a 30-second take costs six times a 5-second one.
4. **Check top-up terms.** kie.ai adds a 10% bonus on larger top-ups. Atlas Cloud's 20% is a time-limited discount. TokLab balances do not expire.

## Calling Seedance 2.5 on TokLab

One request, on the same endpoint as every other video model on TokLab:

```bash
curl https://api.toklab.ai/v1/videos/generations \
  -H "Authorization: Bearer $TOKLAB_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "dreamina-seedance-2-5",
    "content": [{"type": "text", "text": "A red paper boat drifts across a calm pond at sunrise, slow dolly-in"}],
    "resolution": "720p",
    "ratio": "16:9",
    "duration": 5,
    "generate_audio": true
  }'
```

The response is a task ID. Poll `GET /v1/tasks/{id}` until the status is `completed`; there is no completion webhook. The finished task reports the tokens it billed in `output.usage.completion_tokens`.

## FAQ

### What is the cheapest Seedance 2.5 API?

At list prices on November 4, 2026, reapi.ai's Eco tier is the lowest at 720p ($0.201 a second) and 1080p ($0.347). For the standard model at ByteDance's rate, TokLab, BytePlus ModelArk, OpenRouter, Replicate and Runware are within 0.3% of each other: about $0.103 a second at 480p and $0.231 at 720p. Of those five, only OpenRouter lists 4K.

### How much does a 5-second Seedance 2.5 video cost?

At ByteDance's rate, a 16:9 clip costs about $0.51 at 480p, $1.16 at 720p and $2.84 at 1080p. On fal.ai the same clips cost about $1.10, $2.37 and $5.82.

### Does a reference video change the price?

Yes. The rate drops (TokLab charges $6.38 per million tokens instead of $10.681), but the reference video's length is billed on top of the output, so the whole request usually costs more.

### Can I get Seedance 2.5 in 4K?

Among the platforms here, OpenRouter lists 4K at $2.08 a second and WaveSpeedAI at $1.80. BytePlus, TokLab and Replicate stop at 1080p or 720p.

### Why does fal.ai cost twice as much?

fal.ai bills $0.0214 per 1,000 video tokens, which is $21.40 per million, against ByteDance's $10.70. A clip has the same token count everywhere, so the price doubles.

### Is Seedance 2.5 free anywhere?

None of these platforms sells it for free. kie.ai says new users get free credits to test models. TokLab has no free Seedance tier.

## Conclusion

The lowest Seedance 2.5 list price is reapi.ai's Eco tier at 720p and 1080p. For the standard model at ByteDance's token rate, TokLab, BytePlus ModelArk, OpenRouter and Replicate all land at about $0.231 a second at 720p; pick among them on resolution (Replicate stops at 720p, OpenRouter goes to 4K), billing and the other models you need. fal.ai, WaveSpeedAI and kie.ai cost 1.4 to 2 times as much for the same clip. Prices move, so check the [live TokLab rates](https://toklab.ai/pricing) and each platform's page before you commit volume.