Skip to main content

Uncensored — This model declares no content filter. SpicyAPI forwards the request without adding a platform content scan; legal and consent duties remain with the customer.Browse every model at this filtering level
bytedance/seedance-2.0-mini/text-to-video

Seedance 2.0 Mini text-to-video from ByteDance builds the whole frame from a written prompt, with sound in the same pass. Clips run 4 to 15 seconds at 480p or 720p, in seven aspect ratios including adaptive, or let the model pick the length.

From $0.025/s30% below the publisher's list price
Run model

Workbench

Input

Text to video
The only source of what appears on screen. Order it as subject, action, camera, light, sound, one idea per sentence. Wording is not reviewed by us, but Seedance's own layer refuses part of it before rendering — kissing is the reliable trigger, while a whisper at the ear or foreheads touching passes. For a literal result, drive the beat from a first frame on image-to-video instead. at least 1 characters
Output framing, adaptive by default, or one of six fixed ratios from 21:9 down to 9:16. With no image to follow, adaptive lets the model pick the shape the prompt suggests; name a fixed ratio when the placement already dictates one. No effect on what you are billed.
affects price
Output resolution: 480p, or 720p by default. It selects a rate tier, and higher tiers mean bigger files and longer renders. Draft at 480p while the wording is still moving, then raise it once for the take you intend to keep.
Durationaffects price
Length of the finished clip in whole seconds, 4 to 15, default 5, and the billed quantity — the charge scales directly with it. -1 hands the choice to the model: the estimate assumes 15 seconds and the charge follows the length delivered. The extra seconds are room to finish one action, not to fit a second idea in; longer pieces are built by chaining shots. no lower than 4 · no higher than 15
Generate audioaffects price
Whether a native soundtrack is rendered in sync with the picture. On by default; switching it off does not lower the rate in this tier. Turn it off for silent loops and for material you will score in post.
Safety checkerOff

SpicyAPI does not run a safety checker on this model. Your prompt and media go to the model as written, and the model itself decides what comes back.

Checking your account before enabling submission and pricing.

Output

Ready for a new run
Example result
Extend
Video editing
Video upscaling

Related models and next tasks

Missing the model you need?

Tell us what to add and where you would use it. Requests go to the people who decide what gets onboarded next.

Prompt-only clips at the lowest-priced step of the 2.0 line

The Seedance 2.0 Mini text-to-video API turns a written prompt into a 4 to 15 second clip at 480p or 720p, with a soundtrack rendered in the same pass. Its field list matches Seedance 2.0 Fast exactly, down to the seven aspect_ratio values and the 1200 second task timeout. What separates the two tiers is the rate per second, which is why Mini suits work counted in runs of calls.

Open the playground
Prompt-only clips at the lowest-priced step of the 2.0 line
Duration
4 to 15 whole seconds, default 5, or -1 for a length the model picks
Resolution
480p or 720p, 720p by default
Aspect ratio
7 values, adaptive by default
Audio
generate_audio on by default, same rate when it is off
Task timeout
1200 seconds for the task, which is not the clip length
Endpoints
text, image and reference to video, plus video extend
Same contract
The field list matches Seedance 2.0 FastSeedance 2.0 Mini text-to-video publishes the same fields, the same enum values, the same defaults and the same 1200 second task timeout as the Fast tier. A request body written against one runs unchanged against the other, so moving a pipeline between them is a change of model id and nothing else.
Positioning
ByteDance prices the 720p tiers for volumeByteDance's own pricing write-up says the models capped at 720p are priced lower to encourage experimentation and high-volume storyboarding, and it names drafting, A/B testing and rapid iteration as the stages they are for. Of the three Seedance 2.0 tiers published here, Mini carries the lowest rate per second.
Ceiling
720p is where this tier stopsresolution accepts 480p and 720p on every Seedance 2.0 Mini endpoint, where base Seedance 2.0 carries two further steps at 1080p and 4K. Nothing else in the request changes between the two families, so a brief drafted here moves up by editing one field when a delivery needs more pixels.

What the Seedance 2.0 Mini API does well

Every item in a run lands at the same ceiling

Every item in a run lands at the same ceiling

resolution takes 480p or 720p on each call, and the first item in a run gets exactly what the thousandth gets. A run is a list of independent requests, so one item can render at 480p while its neighbour renders at 720p without anything being configured between them. The field is set per call, like every other field here.

Smart duration for prompts that do not share a length

Smart duration for prompts that do not share a length

duration_seconds accepts -1, which hands the length decision to the model inside the 4 to 15 second range. A call sent that way is quoted as if it will return the full 15 seconds, and the charge then follows the length that actually comes back. It earns its place where a run's prompts genuinely differ in pace.

  • A bad item is rerun on its own

    Seedance 2.0 Mini publishes no video-edit endpoint, so nothing repairs a clip once it has rendered. Independent calls at least keep the damage to one item: resubmit the single request that came back wrong and leave the rest of the run alone. A wrong parameter, though, is shared by every item that used it.

  • Six fixed frame shapes, with adaptive as the default

    aspect_ratio defaults to adaptive, so the model chooses the frame that the wording already suggests. The six fixed values are 16:9, 9:16, 4:3, 3:4, 1:1 and 21:9, and naming one keeps every item in a run on the same delivery shape. Framing does not select a rate tier, so the choice stays editorial.

  • One idea, several takes, chosen after the fact

    The workable way to pick a shot here is to send the same brief a few times and choose from what comes back, since each call is quoted and settled on its own. Holding duration_seconds, resolution, aspect_ratio and generate_audio constant across those calls is what makes the comparison about the wording.

  • generate_audio decided per call, not per run

    generate_audio defaults to true and the soundtrack arrives inside the delivered clip. The switch sits in the request body like any other field, so a single run can carry scored items for social and silent items for a site header. The rate per second is the same at both settings.

Who Seedance 2.0 Mini is for

  • Volume, where the rate is the constraint

    Automated pipelines

    Volume, where the rate is the constraint

    A service that submits clips as jobs arrive, dozens or hundreds a day, spends on the rate far more than on any single result. Seedance 2.0 Mini has the lowest rate per second of the three Seedance 2.0 tiers listed here, and its schema never changes between items, so the pipeline stays one request shape in a loop.

  • Several takes before a person picks one

    Prompt testing

    Several takes before a person picks one

    Ten calls carrying the same brief, with duration_seconds, resolution and aspect_ratio identical across all ten, return a spread to choose from. The rate at this tier is what keeps a ten-draw step inside a routine budget, and every draw is an ordinary call with its own task id.

  • One request template applied across a library

    Catalogue and SKU work

    One request template applied across a library

    A product library that needs a short clip per entry runs the same request shape once per item, with only the prompt changing between calls. Since the schema is identical for every item, the work is bookkeeping around a stable request body and a list of task ids.

  • Draft a sequence before any shot is approved

    Storyboards and previsualisation

    Draft a sequence before any shot is approved

    ByteDance names drafting, A/B testing and rapid iteration as the stages the 720p tiers are priced for. A run of rough shots at 480p answers whether a sequence holds together at all, and only the shot that survives that pass needs a higher-resolution rerun on base Seedance 2.0.

How to call the Seedance 2.0 Mini API

  1. Write one request template

    The schema is the same for every item in a run, so build one request body and vary only prompt and whatever else genuinely differs per item. Every other field keeps the value you settled on before the run started.

  2. Fix the length before the run

    duration_seconds is the billed quantity and takes whole seconds from 4 to 15, default 5. Send -1 only where the length is genuinely open, because a call sent that way is quoted as a 15 second clip until the result comes back.

  3. Decide adaptive or a fixed ratio once

    aspect_ratio defaults to adaptive, which follows what the prompt implies. Name one of the six fixed values when every item in the run has the same destination shape, and the framing is settled before a single frame renders.

  4. Sweep at 480p, commit at 720p

    resolution is the field that moves a call between the two rate tiers. A short 480p sweep across the parameter combinations is the cheap way to find the settings worth running over the whole batch at 720p.

  5. Resolve each task on its own

    Every endpoint here is asynchronous, and nothing groups a run: each create call returns one task. Keep the mapping from item to task id at your end, then poll or wait on a callback per item, treating a slow item as unrelated to the rest.

Seedance 2.0 Mini worked examples

  • A clip per item across one product line

    One request template with the prompt swapped per item, resolution held at 720p and duration_seconds at 5, so every item in the run is quoted the same way.

    A ceramic pour-over dripper turning slowly on a plain paper backdrop, soft even studio light, static camera, quiet room tone

  • Several takes on one campaign idea

    The same prompt sent ten times with every other field held constant, which is the cheap way to build a shortlist before a person picks the take worth finishing.

    A paper airplane gliding through a sunlit office, gentle arc, soft motion blur on the wingtip, warm afternoon light

  • A storyboard pass before the final render

    resolution at 480p and duration_seconds at 5 keeps a whole sequence draft cheap while the order of the shots is still moving.

    A hiker crests a ridge at sunrise and stops, wind moving through the grass, wide establishing shot, ambient wind and distant birdsong

Seedance 2.0 Mini API questions

Settle the parameters on a handful of 480p calls, then run the batch at 720p with those values frozen into the template. Change one field at a time while you are sweeping, since two calls carrying the same prompt are two separate draws and a difference has to be attributable to something. Keep duration_seconds at a fixed number wherever the length is known, because -1 is quoted as a 15 second clip until the result arrives, and store a task id for every item as it goes out.

All Seedance models10