Seedance 2.5 API — 30-second clips, 480p to 4K
Seedance 2.5 API pricing: ByteDance's Seedance 2.5 costs $0.1152 per second at 480p, $0.2592 at 720p, $0.6415 at 1080p and $1.955 at 4K on Makernice (price list v12, November 3, 2026). Call it with model id seedance-2.5 for 4–30 second clips from text, a still, first and last frames or up to 16 references; audio is always on.
Highlights
4–30 s · 480p / 720p / 1080p / 4K · Text to video · Image to video · First and last frame · Reference to video
Best for
Longer shots, high-resolution deliverables, and pipelines that anchor a clip with many reference images.
Input
Text, first frame, last frame, up to 16 reference images
Output
Video with audio
Checking your session…
Seedance 2.5 API pricing
Prepaid in US dollars. The hold is converted to a charge when the run completes. If a run fails or times out, the hold is released in full.*
Prices as of November 3, 2026 (price list v12).
Note: if a finished output is blocked by output moderation for a policy violation, the run is still charged. You can contact support for a manual review.
Cost of your first test
$0.46
One 4-second clip at 480p.
What your top-up buys
Top up$15
≈ 32 test runs
$50
≈ 108 test runs
$100
≈ 216 test runs
Top-ups are $15–$1,000. No free tier, no subscription.
Price per clip (price list v12, November 3, 2026; same for all four modes)
| Resolution | Per second | 5 s | 10 s | 15 s | 30 s |
|---|---|---|---|---|---|
| 480p | $0.1152 | $0.58 | $1.15 | $1.73 | $3.46 |
| 720p (default) | $0.2592 | $1.30 | $2.59 | $3.89 | $7.78 |
| 1080p | $0.6415 | $3.21 | $6.42 | $9.62 | $19.25 |
| 4K | $1.955 | $9.78 | $19.55 | $29.33 | $58.65 |
Clip totals are unit price × seconds, rounded to the cent. Your quote shows the exact amount before you run anything. Live rates: Pricing or GET https://api.makernice.com/v1/pricing (no key needed).
| Model | Mode | Resolution | Our price | Official | You save |
|---|---|---|---|---|---|
Seedance 2.5 ByteDance | Text to video480p | 480p | $0.1152/s | ||
Seedance 2.5 ByteDance | Text to video720p | 720p | $0.2592/s | ||
Seedance 2.5 ByteDance | Text to video1080p | 1080p | $0.6415/s | ||
Seedance 2.5 ByteDance | Text to video4K | 4K | $1.955/s | ||
Seedance 2.5 ByteDance | Image to video480p | 480p | $0.1152/s | ||
Seedance 2.5 ByteDance | Image to video720p | 720p | $0.2592/s | ||
Seedance 2.5 ByteDance | Image to video1080p | 1080p | $0.6415/s | ||
Seedance 2.5 ByteDance | Image to video4K | 4K | $1.955/s | ||
Seedance 2.5 ByteDance | First and last frame480p | 480p | $0.1152/s | ||
Seedance 2.5 ByteDance | First and last frame720p | 720p | $0.2592/s | ||
Seedance 2.5 ByteDance | First and last frame1080p | 1080p | $0.6415/s | ||
Seedance 2.5 ByteDance | First and last frame4K | 4K | $1.955/s | ||
Seedance 2.5 ByteDance | Reference to video480p | 480p | $0.1152/s | ||
Seedance 2.5 ByteDance | Reference to video720p | 720p | $0.2592/s | ||
Seedance 2.5 ByteDance | Reference to video1080p | 1080p | $0.6415/s | ||
Seedance 2.5 ByteDance | Reference to video4K | 4K | $1.955/s |
Makernice also offers a separately priced 4K output option; ByteDance's official API tops out at 1080p.
What is the Seedance 2.5 API?
Seedance 2.5 is ByteDance's video generation model built for long takes. It has the highest ceiling in the Seedance family on Makernice: up to 30 seconds per clip and up to 4K. Audio is always on (no toggle). On the Makernice API it's the model id seedance-2.5. It uses the same asynchronous flow as every other model here: quote, create, poll. Moving from Seedance 2.0 or Kling 3.0 means changing one field.
Supported modes and limits
Parameters, limits and billing rules to check before you call it.
Model id
seedance-2.5
Modes
t2v text-to-video · i2v image-to-video · flf first and last frame · r2v reference-to-video
Duration
4–30 seconds, whole seconds (default 5)
Resolution
480p, 720p (default), 1080p, 4k
Aspect ratio (t2v, i2v, flf, r2v)
adaptive, 16:9, 9:16, 1:1, 4:3, 3:4, 21:9, 9:21
Reference images (r2v)
up to 16, in reference_images
Optional fields
seed (0–2,147,483,647)
Audio
Always on (no toggle); sending generate_audio returns 400 invalid_request
Prompt length
1–4,000 characters
Image input
JPEG / PNG / WebP, up to 10 MB each, or an https link. Uploaded images can be used for 6 days.
Output
Video with audio. Files are kept about 30 days, so download them promptly.
Price
From $0.1152/s · Quotes are locked for 12 minutes. Per-resolution prices are in the table above.
Failed runs
Failed, timed-out or canceled-while-queued runs release the hold in full.
From GET /v1/models, October 28, 2026.
4 ways to generate, one endpoint
Set mode, send the inputs it needs — that's it.
Text to video
mode: "t2v" · prompt
Just a prompt — the model builds the shot from scratch.
Image to video
mode: "i2v" · image_url
Send an image in image_url and it becomes the opening frame.
First and last frame
mode: "flf" · image_url + last_image_url
Set the first and last frame; the model animates everything in between.
Reference to video
mode: "r2v" · reference_images (≤ 16)
Add up to 16 reference images to lock in characters, products and style.
What you can build with this model
Workflows this model is a good fit for.
Extended single-take scenes
Generate 20 to 30 second establishing shots or slow camera moves in one request instead of stitching several short clips.
Try it in the playground4K masters for post-production
Render at 4K when the footage is headed for a larger canvas or heavy cropping, and drop to 720p while you iterate.
Try it in the playgroundReference-driven consistency
Pass up to 16 reference images, such as characters, props and style frames, to keep a series visually consistent.
Try it in the playgroundWhy run Seedance 2.5 on Makernice
Quote before you pay
Every run gets an exact quote first, locked for 12 minutes. You're charged that amount — never more.
Failed runs are free
The hold is converted to a charge when the run completes. If a run fails or times out, the hold is released in full, and so does canceling while it's queued. Note: if a finished output is blocked by output moderation for a policy violation, the run is still charged. You can contact support for a manual review.
Switch models with one field
Same request, key, balance and console for all 17 models. Not happy with the result? Change model and run it again.
Prepaid, no subscription
Top up a US-dollar balance and pay per run. No monthly fee, no subscription.
Your balance never expires
Topped-up balance stays in your account indefinitely. No inactivity fees, no decay.
Automatic failover
If an upstream channel goes down, your request is routed to a backup channel automatically — no code changes on your side.
Which should you pick: Seedance 2.5, Seedance 2.0, Seedance 2.0 Fast?
Compare modes, length, resolution and references, then run the same prompt on each.
| Seedance 2.5 | Seedance 2.0 | Seedance 2.0 Fast | |
|---|---|---|---|
| Modes | Text to video · Image to video · First and last frame · Reference to video | Text to video · Image to video · First and last frame · Reference to video | Text to video · Image to video · First and last frame · Reference to video |
| Duration | 4–30 s | 4–15 s | 4–15 s |
| Resolution | 480p / 720p / 1080p / 4K | 720p / 1080p | 720p (single tier) |
| Reference images | Up to 16 | Up to 9 | Up to 9 |
| Audio | Always on (no toggle) | Always on (no toggle) | Always on (no toggle) |
| Starting price | $0.1152/s | $0.1704/s | $0.1027/s |
Short version: Seedance 2.5 for clips over 15 seconds, 4K or more than 9 references. Seedance 2.0 for 720p/1080p clips up to 15 seconds. Seedance 2.0 Fast (fixed 720p, $0.1027/s) for drafts. Full comparison →
Integrate Seedance 2.5 in three steps
Quote, hold, charge or release — each step shows how billing works.
Create an API key
Sign in, top up and create a key in the console. Set a spend limit per key so nothing runs away.
// Console → API keys
Quote and create
POST the input to /v1/generations/quote with the model field set to seedance-2.5. The price is locked for 12 minutes; creating the generation with the quote id holds that amount from your balance.
// POST /v1/generations/quote// POST /v1/generations
Poll the result
GET /v1/generations/{id} until the status is succeeded, then download the output. You're charged the quoted amount on success; failed, timed-out or canceled-while-queued runs release the hold in full.
// GET /v1/generations/{id}
Retrying with the same Idempotency-Key returns the same generation instead of charging twice. More: How AI video API pricing works.
Quickstart: quote → create → poll
curl
# 1) Quote: returns an id and the exact amount (in millionths of a USD)
curl https://api.makernice.com/v1/generations/quote \
-H "Authorization: Bearer $MAKERNICE_API_KEY" \
-H "Content-Type: application/json" \
-d '{"model":"seedance-2.5","mode":"t2v","prompt":"A lone cyclist crosses a rain-soaked neon street at night, reflections shimmering on the asphalt, camera tracking alongside.","duration":10,"resolution":"720p","aspect_ratio":"16:9"}'
# 2) Create: same input + quote_id, with an Idempotency-Key
curl https://api.makernice.com/v1/generations \
-H "Authorization: Bearer $MAKERNICE_API_KEY" \
-H "Content-Type: application/json" \
-H "Idempotency-Key: seedance25-demo-001" \
-d '{"quote_id":"qt_…","input":{"model":"seedance-2.5","mode":"t2v","prompt":"A lone cyclist crosses a rain-soaked neon street at night, reflections shimmering on the asphalt, camera tracking alongside.","duration":10,"resolution":"720p","aspect_ratio":"16:9"}}'
# 3) Poll until status is "succeeded", then download
curl https://api.makernice.com/v1/generations/gen_… -H "Authorization: Bearer $MAKERNICE_API_KEY"
curl -L -o seedance-2-5.mp4 https://api.makernice.com/v1/generations/gen_…/outputs/0 \
-H "Authorization: Bearer $MAKERNICE_API_KEY"Python (requests)
import os, time, uuid, requests
API = "https://api.makernice.com/v1"
H = {"Authorization": f"Bearer {os.environ['MAKERNICE_API_KEY']}"}
job = {
"model": "seedance-2.5",
"mode": "t2v",
"prompt": "A lone cyclist crosses a rain-soaked neon street at night, "
"reflections shimmering on the asphalt, camera tracking alongside.",
"duration": 10,
"resolution": "720p",
"aspect_ratio": "16:9",
}
quote = requests.post(f"{API}/generations/quote", json=job, headers=H, timeout=30).json()
print(f"Quoted ${int(quote['amount']) / 1e6:.2f}, valid until {quote['expires_at']}")
gen = requests.post(
f"{API}/generations",
json={"quote_id": quote["id"], "input": job},
headers={**H, "Idempotency-Key": str(uuid.uuid4())},
timeout=30,
).json()
while gen["status"] not in ("succeeded", "failed", "canceled"):
time.sleep(5)
gen = requests.get(f"{API}/generations/{gen['id']}", headers=H, timeout=15).json()
if gen["status"] == "succeeded":
video = requests.get(f"https://api.makernice.com{gen['outputs'][0]['url']}", headers=H, timeout=120)
open("seedance-2-5.mp4", "wb").write(video.content)
else:
print("Not charged:", gen["status"]) # the hold is released in fullSwitch models in one line
- "model": "seedance-2.5",
+ "model": "kling-3.0-pro",Each model has its own limits. Seedance 2.0, for example, accepts 4–15 seconds at 720p or 1080p, and a request outside a model's range returns 400 invalid_request before anything is held. Check GET /v1/models.
Docs/API/SEEDANCE-2-5
API reference
Copy-paste requests for this model. The full parameter reference is in the docs.
curl https://api.makernice.com/v1/generations/quote \
-H "Authorization: Bearer $MAKERNICE_API_KEY" \
-H "Content-Type: application/json" \
-d '{
"model": "seedance-2.5",
"mode": "t2v",
"prompt": "A slow aerial push-in over a misty bamboo forest at sunrise, soft golden light filtering through the leaves, cinematic and calm.",
"duration": 5,
"resolution": "720p"
}'Invalid request
400 invalid_request names each failing field. Fields the mode doesn't use are rejected too. Nothing is held.
Key or balance
401 means a missing or invalid key; 402 means the balance or the key's spend limit isn't enough.
Quote problems
A changed price-affecting field returns 422 quote_mismatch; an expired quote 410; a used one 409. Get a new quote.
Failed generation
A failed generation is never charged — the hold is released in full. Note: if a finished output is blocked by output moderation for a policy violation, the run is still charged. You can contact support for a manual review. Keep the generation id and retry when you're ready.
Frequently asked questions
Common questions about this model.
What is the Seedance 2.5 API?
seedance-2.5, using one API key and one prepaid balance shared with 16 other models. It makes 4–30 second clips at 480p to 4K from text, an image, first and last frames, or up to 16 reference images. Audio is always on (no toggle).How much does the Seedance 2.5 API cost per second?
POST /v1/generations/quote for the exact amount before you run it. Note: if a finished output is blocked by output moderation for a policy violation, the run is still charged. You can contact support for a manual review.Is Seedance 2.5 free?
How long can a Seedance 2.5 video be?
Does Seedance 2.5 support 4K?
How many reference images can I send?
reference_images, using reference-to-video mode (r2v).Am I charged if a generation fails?
Does Seedance 2.5 generate audio?
When did the Seedance 2.5 API become available?
seedance-2.5.Can I switch between Seedance 2.5 and Seedance 2.0 with the same code?
model field. Seedance 2.0 accepts 4–15 seconds at 720p or 1080p with up to 9 references, so adjust duration and resolution if needed.Is Seedance 2.5 worth it?
model field.Prompts for Seedance 2.5
Browse all promptsSleep-Deprived Mom Mix-Ups: 3D Family Comedy
A 30-second vertical 3D comedy prompt about an exhausted young mother whose mix-ups pile up with a baby on her hip: dishes in the washing machine, keys in her husband's tea, an egg cracked into the bin, an omelette on his head. Each beat follows the finished film, for Seedance 2.5 text-to-video.
Seedance 2.5
Betrayal and a Slap on a Subway Platform
A 30-second text-to-video prompt that generates a live-action melodrama on a modern subway platform: a young woman with coffee and phone catches her boyfriend kissing someone else, drops both, confronts and slaps him, then walks away in tears. All three characters are written out in words, following the clip.
Seedance 2.5
A Red-Haired Girl Between Pink and Green Dragons
A 30-second photoreal fantasy prompt that generates a red-haired young woman cuddling a giant pink dragon in a flowered meadow, meeting a green dragon that walks out of the forest, and standing between both as they spread their wings at sunset. It is rebuilt shot by shot from the finished clip.
Seedance 2.5
A Girl in a Purple Cloak Meets a Giant Robot
A 25-second photoreal fantasy prompt that generates a little girl in a purple hooded cloak playing with mechanical spiders at a rainy café until a giant blue-eyed robot crouches before her, opens its hand on a glowing crystal and receives her white daisy. It is rewritten from the finished clip.
Seedance 2.5
More Posts
Related
Related models
Compare all modelsSimilar models worth comparing.
VideoSeedance 2.0
The all-rounder for ads and social: 4–15 s clips at 720p or 1080p from text, an image, first and last frames, or 9 references. Audio is always on (no toggle).
VideoSeedance 2.0 Fast
-15%Seedance 2.0 at one 720p tier and a lower per-second price: same modes, same 4–15 seconds. Audio is always on (no toggle). Draft here, finish on 2.0 or 2.5.
VideoMiniMax H3
-25%MiniMax's H3 with a 2K default and a 4K option — 5–15 second clips from text, a still, first and last frames, or up to 5 references.
VideoMiniMax H3 Max
-9%Draft at 480p, finish at 1080p: H3 Max gives you three resolution tiers, 5–15 second clips and up to 4 reference images.
● Start building
See what it does with your prompt
Test it in the playground, or grab an API key and ship it today.
- Quote before you pay
- Failed runs are free
- Prepaid, no subscription
- Balance never expires