How resolution tiers shape what Seedance 2.5 costs

Updated 2026-10-02

Ask what Seedance 2.5 costs and the honest answer has three parts: which host, how many seconds, and which resolution tier. The tier is the part people skip, and it is often the largest lever you control after choosing the variant. This page explains how tiers show up in pricing and how to read the live tables without comparing things that are not comparable.

Price is per second, per tier, per host

Each host sets its own rate for each tier it offers. A single "Seedance 2.5 price" therefore does not exist; there is a grid of host by tier. VideoRouter bills once at job creation: the host's per-second rate for the tier you request, multiplied by the requested (snapped) duration, plus a 2% platform fee on image and video usage. Polling is free and jobs that fail upstream are not billed. The tier you request is part of the request body, via the resolution field.

The live table below compares the cheapest and priciest host for each Seedance row, and each row states which tier it compares. Read that tier label before you read the number:

Per second, before VideoRouter's 2% platform fee. For tiered models each row compares the resolution tier with the widest host-to-host gap. Built 2026-10-02 from the live catalog.

The tier names on Seedance 2.5

The Seedance 2.5 model page lists a family of tier strings rather than a single resolution ladder:

• Plain tiers: 480p, 720p, 1080p and 2160p.
• -sr variants: 720p-sr, 1080p-sr and 1440p-sr.
• -esr variants: 720p-esr, 1080p-esr, 1440p-esr and 4k-esr, with a 60 fps option shown for one of them.

VideoRouter's documentation describes the suffixed tiers as enhanced-resolution and super-resolution variants that some hosts list and others do not. The practical meaning for pricing is not that a suffixed tier costs a fixed multiple of its plain sibling. It is that each is a separate priced item, and a host that does not carry a tier cannot serve it at any price. Do not assume that a host's low base-tier rate means it can serve your chosen suffixed tier.

Compare like with like

The most common reading mistake is comparing a 480p rate on one host with a 720p rate on another. Three rules keep a comparison honest:

1. Match the tier string exactly. 720p and 720p-sr are different products. Use the tier label printed in the table row.
2. Match the variant. Seedance 2.5, 2.0, Fast, Mini and 1.5 Pro are separate rows with separate rates. A host being cheap on Mini says nothing about it on 2.5.
3. Check that the host offers the tier at all. A table that shows the cheapest host at a base tier does not tell you the same host serves your -esr tier.

The table's "tier" column is the tier the comparison was made at. When your target tier differs, open the model page for the host-by-tier detail instead of extrapolating from the row.

Requesting a tier, and what happens if you get it wrong

{
  "model": "bytedance/seedance-2.5",
  "prompt": "a paper airplane gliding over a city",
  "duration_secs": 5,
  "resolution": "720p",
  "aspect_ratio": "16:9"
}

An unsupported resolution and aspect_ratio combination never returns a 400. It is ignored and the model default is used, as if you had omitted both. That has a billing implication: if a mistyped tier falls back to a default you did not intend, you pay the default's rate, not the one you budgeted. Log the tier you requested next to each job, and in tests inspect the returned file's actual dimensions with ffprobe before trusting a batch.

A tier decision framework

1. Start from the delivery target. A vertical social clip rarely needs the top tier; a large-screen master might. Write the minimum tier that satisfies the real destination.
2. Draft one step below. Judge motion and prompt adherence at a low tier on a lighter variant, then re-run approved shots at the delivery tier. A draft predicts intent, not exact pixels.
3. Treat suffixed tiers as a deliberate purchase. Use them only for clips that will actually be seen at that quality, and only after you have confirmed on the model page which hosts serve them.
4. Re-read the table before each project. Rates and host lineups change; a number remembered from last month is not a quote.

Turning a tier rate into a clip cost

Because billing is requested seconds times rate, doubling clip length doubles cost at a given tier, and raising the tier changes the rate on every second. The cost view computes 5 second, 10 second and 60 second clips at the cheapest host for each row:

Computed from each model's cheapest current host per resolution tier, before VideoRouter's 2% platform fee. Built 2026-10-02 from live data. Failed upstream jobs are not billed; re-generations are, so multiply by your retry rate.

Those are floor figures at the tier the table states. A project that renders its drafts at a higher tier than necessary pays the higher rate on every discarded clip, which is why tier discipline in the exploration stage matters more than shaving a fraction off the final render. The per-minute budget guide covers the retry multiplier that sits on top.

Where to go next

The pricing page and cost page show the full current figures, and the calculator lets you enter your own clip lengths. To verify a tier on real output, create a key and run one short request per tier you are considering, then compare the files. The quickstart has a working request.

Frequently asked questions

Keep reading

• How to Reduce Your Seedance API Cost — Five Practical Levers
• Seedance API Cost Per Minute: Budget Formula & Retry Math
• Seedance vs Kling API Pricing: A Fair Cost Comparison
• Seedance 2.0 Mini vs Fast vs Full: A Cost Strategy for Teams