How much does the Seedance 2.0 API cost? Pricing is billed per second of generated video (at 720p) on fal.ai. Audio generation is free — the cost is the same whether generate_audio is true or false.
| Endpoint | 720p / sec | Per 1K tokens |
|---|---|---|
bytedance/seedance-2.0/text-to-video |
$0.3034 | $0.0140 |
bytedance/seedance-2.0/image-to-video |
$0.3024 | $0.0140 |
bytedance/seedance-2.0/reference-to-video (images only) |
$0.3024 | $0.0140 |
bytedance/seedance-2.0/reference-to-video (with video inputs) |
$0.1814 | $0.0140 |
bytedance/seedance-2.0/fast/text-to-video |
$0.2419 | $0.0112 |
bytedance/seedance-2.0/fast/image-to-video |
$0.2419 | $0.0112 |
bytedance/seedance-2.0/fast/reference-to-video (images only) |
$0.2419 | $0.0112 |
bytedance/seedance-2.0/fast/reference-to-video (with video inputs) |
$0.14515 | $0.0112 |
Tokens are computed as:
tokens = (height * width * duration * 24) / 1024
For reference-to-video, the input video duration also counts toward tokens, and if a video reference is supplied the final price is multiplied by 0.6 (a ~40% discount).
| Scenario | Duration | Est. cost |
|---|---|---|
| Text-to-video, 6 s | 6 s | ~$1.45 |
| Image-to-video, 5 s | 5 s | ~$1.21 |
| Reference-to-video with 1 image, 8 s | 8 s | ~$1.93 |
| Reference-to-video with video input, 8 s | 8 s | ~$1.16 |
Prices are illustrative. Confirm the latest numbers on the fal.ai pricing page.
- Prefer the fast tier for iteration; move to standard only for final renders.
- Use 480p while prompt-tuning, then re-render the winner at 720p.
- Use the reference-to-video endpoint with a short video reference — the 0.6x multiplier is the cheapest way to run Seedance 2.0.
- Cache results by
seed— re-rendering with the same seed + prompt gives near-identical outputs.