If you came here for a ranked table of Seedance 2.5 prices per video, the honest answer is that no provider has published a comparable per-video Seedance 2.5 price at the time of writing. What follows is the thing that actually helps you plan a budget: how Seedance video cost is constructed, which two billing models you will meet, and real arithmetic on the Seedance tiers that are live today.
Key Takeaways
- No provider has published a comparable per-video Seedance 2.5 price at the time of writing, so any ranked "cheapest 2.5" table you find elsewhere is either extrapolated or invented.
- Seedance 2.5 is live on first-party ByteDance channels (Volcano Engine Ark in China, BytePlus ModelArk internationally). Aggregators are in prelaunch or Day-0-commitment state.
- Cost per video is driven by four variables: output duration, output resolution, output frame rate, and whether the input includes a video.
- Two billing models collide here. First-party ByteDance channels bill by token with minimum-token floors when the input includes video. Atlas Cloud bills a flat rate per second of output duration.
- You can build a defensible budget today by anchoring on the live Seedance tiers on Atlas Cloud: Seedance 2.0 at $0.112/s, Seedance 2.0 Fast at $0.09/s, Seedance 2.0 Mini at $0.056/s, and Seedance 1.5 Pro at $0.047/s, all billed on output duration.
- Atlas Cloud has published a Seedance 2.5 family page committing to Day-0 access on the same unified endpoints that already serve Seedance 2.0 and 1.5. That listing is currently marked Coming Soon and no Atlas Cloud price for 2.5 has been published yet.
Why nobody can hand you a cheapest 2.5 table yet
Seedance 2.5 is ByteDance's next-generation audio-video joint generation model, built for 30-second storytelling. It generates up to 30 seconds in a single generation with the option to extend twice, has smoother motion, understands the intention, framing and cinematic language of a reference video rather than just copying its motion, and adds production capabilities such as white-model control and green-screen editing.
Where it is reachable is narrower than the chatter suggests. Verified at the time of writing:
- Volcano Engine Ark, the China-region console, has it live with a published pricing section.
- BytePlus ModelArk, the international console, documents pricing and publishes a Seedance 2.5 Series Pricing Calculator.
- WaveSpeedAI has an explicit prelaunch page telling users to use Seedance 2.0 endpoints today.
- Replicate returns a 404 for its Seedance 2.5 path.
- fal.ai could not be verified either way (rate-limited), so this article makes no claim about it.
- Atlas Cloud has a Seedance 2.5 family page marked Coming Soon, with a Day-0 commitment and no published price.
The first-party tables that do exist are token-rate tables gated behind a calculator, split into "input without video" and "input with video". They do not yield a single attributable per-video figure, which is exactly why a credible cheapest-provider ranking for 2.5 does not exist yet. Anyone publishing one is filling in blanks.
What actually drives cost per video
Strip away the branding and a Seedance render's cost is a function of four things.
- Output duration. The single biggest lever. Every billing model in this space scales with how long the finished clip is.
- Output resolution. Width times height enters the token formula directly, so 1080p costs materially more than 720p at the same length.
- Output frame rate. A 30 fps render consumes more than a 24 fps render at identical length and resolution.
- Whether the input includes video. The variable most teams forget, and the one that breaks naive estimates.
That last point matters. On first-party ByteDance channels, estimated token consumption is (input video duration + output video duration) x output width x output height x output frame rate / 1024. Your input video duration sits inside the formula, so feeding a 20-second reference into a 10-second render is not the same cost as a text prompt into a 10-second render.
Token billing with floors versus flat per-second billing
Estimated video price equals token unit price multiplied by token consumption, estimated with the formula above. Some models support a draft mode for fast, low-quality test renders at lower token consumption via a conversion factor. Whatever the estimate says, the authoritative number is the usage.completion_tokens field returned by the API, so first-party billing is inherently a post-hoc reconciliation.
Then the floor. For the Seedance 2.0 series and for Seedance 2.5, when the input includes video, a minimum token consumption floor applies: if your estimate lands below the floor, the floor is billed anyway. The floor depends on resolution, aspect ratio and output duration. BytePlus also documents that within its listed range the lowest price corresponds to 2 to 4 seconds of input and the highest to 30 seconds of input.
Practically, a short with-video render can cost more than the formula suggests, and short-clip-heavy workloads are the most exposed. Flat per-second billing works differently: Atlas Cloud bills video by output duration at a published per-second rate, shown live next to the Run button in the Playground before you spend anything. Input video duration does not enter the rate, and there is no floor to reconcile afterwards.
| Billing dimension | First-party ByteDance channels | Atlas Cloud | Typical aggregator per-second pricing |
|---|---|---|---|
| Unit of billing | Tokens | Seconds of output duration | Seconds of output duration |
| Cost formula inputs | Input video duration, output duration, width, height, frame rate | Output duration | Output duration |
| Minimum floor when input includes video | Yes, varies by resolution, aspect ratio and duration | Not applicable | Not applicable |
| Estimation before you run | Pricing calculator | Live unit price in the Playground | Published rate card |
| Reconciliation after you run | Yes, from usage.completion_tokens | Not required | Not required |
| Draft mode discount | Supported on some models | Not applicable | Varies by provider |
| Published Seedance 2.5 price | Token-rate tables behind a calculator | None yet, listing marked Coming Soon | None comparable published |
For a team budgeting a 30-second render, per-second-on-output is arithmetic you can do in your head. Token-with-floors needs a calculator up front and a reconciliation afterwards. Neither is wrong, but they demand very different finance workflows, and that difference is a legitimate part of choosing a provider.
Worked cost per video on the live Seedance tiers
The numbers below are the live Atlas Cloud rates for Seedance 2.0, Seedance 2.0 Mini and Seedance 1.5 Pro. They are not Seedance 2.5 rates and should not be read as a forecast of one. They are useful because they are real, billed on output duration, and let you build a budget model you can plug a future rate into.
Take a 10-second clip, billed on output duration:
- Seedance 2.0 at $0.112/s: 10 x 0.112 = $1.12 per video. The Fast variant at $0.09/s makes the same clip $0.90.
- Seedance 2.0 Mini at $0.056/s: 10 x 0.056 = $0.56 per video, the cheapest Seedance 2.0 tier on Atlas Cloud.
- Seedance 1.5 Pro at $0.047/s: 10 x 0.047 = $0.47 per video, or $0.18 on the Fast variant at $0.018/s, with native synchronized audio, multilingual lip sync, and a 4 to 12 second clip range.
Scale that to a 100-video batch of 10-second clips and the spread is $112 on Seedance 2.0, $90 on Seedance 2.0 Fast, $56 on Seedance 2.0 Mini, and $47 on Seedance 1.5 Pro. A single 30-second clip on Seedance 2.0 lands at 30 x 0.112 = $3.36.
Now run the same shape through the token formula. A 10-second 1280x720 output at 24 fps with no video input estimates as (0 + 10) x 1280 x 720 x 24 / 1024 = 216,000 tokens. Add a 10-second video input and the first term doubles to 20, so the estimate doubles to 432,000 tokens before any floor is applied. Multiply by whatever token unit price your channel publishes and you have your estimate. That is the mechanism. The unit price for 2.5 is the piece this article deliberately does not invent.
For a cross-provider sanity check on how much per-second rates diverge, here is a verified Seedance 2.0 comparison at 720P with video input. Again, this is a Seedance 2.0 spec comparison, not a 2.5 one.
| Provider | Seedance 2.0, 720P, with video input |
|---|---|
| Kie.ai | $0.125 |
| Atlas Cloud | $0.1486 |
| WaveSpeed | $0.15 |
| OpenRouter | $0.1512 |
| Fal.ai | $0.1814 |
The spread from cheapest to most expensive on the same model and the same spec is roughly 45 percent. That is the order of magnitude of provider-choice savings to expect once 2.5 rates land, and it is a far better planning assumption than a made-up number.
Budgeting for 2.5 before the price exists
A defensible budget today looks like this.
First, define your unit of work precisely: duration, resolution, frame rate, and whether the input includes video. Without those four you do not have a cost question, you have a wish.
Second, anchor on a live tier. If your workload resembles a Seedance 2.0 job, budget at $0.112/s of output, or $0.09/s on the Fast variant. If cost is the binding constraint, Seedance 2.0 Mini at $0.056/s is the cheapest Seedance 2.0 tier on Atlas Cloud, and Seedance 1.5 Pro at $0.047/s, or $0.018/s on its Fast variant, covers short clips with native synchronized audio.
Third, sanity-check against the wider live video catalog on the same key. Grok Imagine Video v1.5 sits at $0.08/s, Youchuan V8.2 at $0.086/s, Kling V3.0 Turbo at $0.112 standard and $0.095 discounted, Wan 2.7 Spicy at $0.10, Gemini Omni Flash text-to-video at $0.125, and MiniMax H3 at $0.14. A new flagship generally does not price below that band, which gives you a floor for scenario planning.
Fourth, hold a contingency for the with-video floor if any part of your pipeline sends reference video to a first-party channel. Short renders with video input are where estimates most often understate the bill.
Atlas Cloud is a full-modal AI inference platform that serves 300+ curated models across text, image and video under one API key and one bill. Atlas Cloud publishes a live per-model unit price next to the Run button in the Playground, so the price you budget against is the price you see before you spend. Atlas Cloud holds SOC II certification and is HIPAA compliant, with encryption at rest and in transit, and operates with no deposit, no minimum commitment and no waitlist.
What to build against today
Build the integration now against a Seedance tier that is live, then change one string later. Seedance 2.0, Seedance 2.0 Mini and Seedance 1.5 Pro are all reachable today on the same Atlas Cloud key at atlascloud.ai/models/seedance2, with rates at atlascloud.ai/pricing. The Seedance 2.5 family page at atlascloud.ai/models/seedance-2.5 states that Atlas Cloud will bring Day-0 access on the same unified endpoints that already serve Seedance 2.0 and 1.5, and is currently marked Coming Soon.
Because the endpoint contract does not change, your retry logic, cost logging and queueing all survive the switch. For text models, if you already run an OpenAI SDK application, the integration work is a base_url and key change; video is a separate submit-and-poll REST flow on the same key, documented at atlascloud.ai/docs. For teams outside China, first-party access routes to BytePlus ModelArk rather than the China-region Volcano Engine Ark console, which is a routing consideration rather than a pricing one.
FAQ
Q: Which provider has the cheapest Seedance 2.5 API right now? A: Nobody can answer that honestly yet, because no provider has published a comparable per-video Seedance 2.5 price at the time of writing. First-party channels publish token-rate tables behind a pricing calculator, and aggregators are in prelaunch or Day-0-commitment state.
Q: How do I estimate cost per video for a Seedance render? A: On first-party ByteDance channels, estimated token consumption is (input video duration + output video duration) x output width x output height x output frame rate / 1024, multiplied by the token unit price, with a minimum token floor when the input includes video. On Atlas Cloud, video is billed flat per second of output duration.
Q: What is the cheapest Seedance 2.0 tier I can actually call today? A: Seedance 2.0 Mini at $0.056/s is the cheapest Seedance 2.0 tier on Atlas Cloud for text-to-video, image-to-video and reference-to-video. Seedance 1.5 Pro is $0.047/s, or $0.018/s on its Fast variant for 4 to 12 second clips with native synchronized audio.
Q: Why does a short clip sometimes cost more than my estimate? A: Because of the minimum token consumption floor that applies on the Seedance 2.0 series and Seedance 2.5 when the input includes video. If your estimate falls below the floor, the floor is billed. The floor depends on resolution, aspect ratio and output duration.
Q: Will my code need rewriting when Seedance 2.5 becomes available on Atlas Cloud? A: The published family page commits to Day-0 access on the same unified endpoints that already serve Seedance 2.0 and 1.5, so the change is the model identifier rather than the integration. No Atlas Cloud price for 2.5 has been published yet.
The bottom line
The cheapest Seedance 2.5 provider cannot be named today, and a table that names one is fabricating it. What can be established is the cost structure: duration, resolution, frame rate and video input drive the number, token billing adds floors and post-hoc reconciliation, and flat per-second billing on output duration makes the arithmetic predictable in advance. Atlas Cloud lists Seedance 2.0 at $0.112/s, Seedance 2.0 Fast at $0.09/s, Seedance 2.0 Mini at $0.056/s and Seedance 1.5 Pro at $0.047/s on a single key alongside 300+ other models, and has published a Seedance 2.5 family page marked Coming Soon with a Day-0 access commitment and no price attached.
