How Video Token Pricing Actually Works: Seedance 2 and Seedance 2 Fast
Understand exactly how Seedance 2 and Seedance 2 Fast pricing works: tokens, resolution, video-in vs text-to-video rates, and what you pay per second.
When I first looked at how Seedance 2 is priced, my immediate reaction was: why isn't this a flat per-video cost? Most image models charge per image. Why are video models different?
The answer comes down to tokens. And once you understand how video tokens work, the pricing makes complete sense.
What Are Video Tokens?
Video tokens aren't like LLM tokens. You're not paying per word or per character. Video tokens are a function of the visual information in your clip: specifically, the number of pixels being processed per second.
For Seedance 2, the formula is straightforward: the more pixels in each frame, and the more frames per second, the more tokens your generation consumes. Seedance 2 runs at 24 FPS. A 16:9 480p clip (864x496 pixels) uses about 10,044 tokens per second of generated video. Scale that up to 720p (1280x720) and you're at roughly 21,600 tokens per second. More than double, because you're processing more than twice the pixel information per frame.
The Pricing Structure
Seedance 2 is priced per 1,000 tokens. There are two unit costs depending on your input type:
| Input Type | Seedance 2 | Seedance 2 Fast |
|---|---|---|
| Text-to-video | $0.0084 / 1K tokens | $0.0067 / 1K tokens |
| Video-to-video | $0.0052 / 1K tokens | $0.0040 / 1K tokens |
The video-in rate is lower because when you provide a reference video as input, the model is doing less heavy lifting from scratch. You're guiding it rather than generating from a blank canvas.
What You Actually Pay Per Second
Here's what the token math translates to in real terms:
Seedance 2 (Standard)
| Resolution | Aspect Ratio | Text-to-Video / s | Video-to-Video / s |
|---|---|---|---|
| 480p | 16:9 | $0.084 | $0.052 |
| 480p | 1:1 | $0.081 | $0.050 |
| 720p | 16:9 | $0.181 | $0.111 |
Seedance 2 Fast
| Resolution | Aspect Ratio | Text-to-Video / s | Video-to-Video / s |
|---|---|---|---|
| 480p | 16:9 | $0.067 | $0.040 |
| 480p | 1:1 | $0.065 | $0.038 |
| 720p | 16:9 | $0.145 | $0.086 |
Seedance 2 vs Seedance 2 Fast: Which One to Use?
Seedance 2 Fast is roughly 20% cheaper. The tradeoff is speed vs. output fidelity. Fast is optimized for quicker inference. Standard Seedance 2 prioritizes quality.
For a typical 5-second clip at 720p 16:9, text-to-video, the math is: Seedance 2 runs about $0.91 and Seedance 2 Fast runs about $0.73. If you're generating 50 clips for an A/B campaign, that 20% difference adds up fast.
My approach: use Seedance 2 Fast for iteration and prototyping, switch to standard Seedance 2 for final production renders where quality matters.
Resolution is Your Biggest Cost Lever
Aspect ratio changes token counts slightly at the same resolution, but not dramatically. The real cost driver is moving between resolutions. Going from 480p to 720p more than doubles your cost, because pixel count more than doubles.
For social content where 480p is perfectly fine (Instagram Reels, TikTok, YouTube Shorts previews), stick to 480p. Reserve 720p for client deliverables, landing page hero videos, or anything where sharpness is non-negotiable.
Audio Is Free
Worth calling out: generating audio alongside your video does not add to the cost. Audio generation is included at no extra charge. If you're building social or ad content, always enable audio. There's no reason not to.
Quick Reference
For the most common workflows, here's what to budget per second of output:
- 480p 16:9, text-to-video, Seedance 2: $0.084/s
- 480p 16:9, text-to-video, Seedance 2 Fast: $0.067/s
- 720p 16:9, text-to-video, Seedance 2: $0.181/s
- 720p 16:9, text-to-video, Seedance 2 Fast: $0.145/s
A 5-second 720p clip costs between $0.73 and $0.91 depending on which variant you use. A 5-second 480p clip lands between $0.34 and $0.42. Build that into your API cost model and there are no surprises on your bill.
You can start generating with Seedance 2 at segmind.com/models/seedance-2.