Seedance 2.0 vs Fast vs Mini: Which Tier? (2026)

Already picked Seedance over Wan? Now pick the tier. ByteDance ships Seedance 2.0 in three flavors on ofox, and they are not three quality grades. They are three resolution ceilings on the same model family, each priced by the resolution you actually request.

Here is the trap the catalog sets: the “from” rates you see next to each tier — $0.04, $0.06, $0.07 — are 480p floor prices, and almost nobody ships 480p. Price scales with resolution. At 720p, where Mini and Fast actually run, Mini is $0.08/s and Fast is $0.13/s. The flagship is the only tier above 720p, at $0.34/s for 1080p and $1.37/s for 4K. All three do audio, image-to-video, and video-to-video, share a 4-second minimum clip, and sit behind one POST /v1/videos call. If you are still deciding Seedance against Alibaba’s Wan, that is the wrong page: read the Seedance 2.0 vs Wan comparison first, then come back here to pick the tier.

TL;DR: Which Tier Should You Pick?

TL;DR: Seedance 2.0’s three tiers are priced by resolution, not by a flat per-tier rate. The headline $0.04 / $0.06 / $0.07 are 480p floors. At 720p — where Mini and Fast actually run — Mini is $0.08/s and Fast is $0.13/s, so Fast is a $0.05/s (62.5%) premium at the same 720p ceiling. The flagship is the only tier that reaches 1080p ($0.34/s) and 4K ($1.37/s). Pick by output resolution first, then by budget, and test Fast against Mini on your own prompts before you pay that premium.

The decision collapses to two questions. What resolution does your delivery format actually need, and how much of your output is disposable? If your clips land on a feed that downscales to 720p or below, Mini clears the requirement at $0.08/s, the lowest 720p rate on the catalog. If you master at 1080p or archive at 4K, only the flagship can produce the pixels, at $0.34/s and $1.37/s. Fast sits in the middle as a 720p tier at $0.13/s, and its whole reason to exist is output quality that a rate card cannot promise.

Everything below is the arithmetic and the caps behind that table, plus the exact code to run all three tiers on one call.

Quick Specs: The Three Tiers Side by Side

Every number here comes from the live ofox model catalog on 2026-07-20. The three tiers share almost every capability. What separates them is the resolution ceiling and a per-second rate that climbs with resolution. Read the table with that in mind, because it is easy to assume “Mini” means “worse motion” when the datasheet only documents a lower resolution cap and a lower price.

Two things carry the whole decision. The Max resolution row is a hard gate: Fast and Mini physically cannot emit 1080p or 4K, so any format that needs those pixels forces the flagship regardless of budget. The price rows show why the tier names mislead. The $0.04 / $0.06 / $0.07 you first see are 480p floors; every real resolution costs more. At 720p, Mini is $0.08/s — half the flagship’s 720p rate and about 38% under Fast. Above 720p there is no choice at all: the flagship is the only tier, at $0.34/s for 1080p and $1.37/s for 4K, roughly 5× to 20× its own 480p floor.

Notice what the table does not say. There is no “quality score” column, because the catalog does not publish one, and neither ByteDance nor ofox documents Fast or Mini as producing lower-fidelity motion at matched resolution. The tier names imply a quality staircase; the datasheet only guarantees a resolution and price staircase. That gap is exactly why the Fast-versus-Mini question needs a real test rather than a spec read, and it is the honest limit of this comparison.

One freshness note, so you are not surprised: ByteDance previewed a Seedance 2.5 flagship in mid-2026 with native 4K and longer single-shot clips. As of 2026-07-20 the three 2.0 tiers here are the ones live on ofox. When 2.5 arrives on the same /v1/videos endpoint, moving to it is the same one-string change as switching between these three.

The Caps That Actually Constrain You

A tier decision is really a decision about two ceilings you cannot argue with: resolution and minimum duration. Get either wrong and the API returns a 400 before you have spent a cent, so it pays to know them cold.

Resolution is the gate that forces the flagship. Fast and Mini top out at 720p. There is no request parameter that lifts that ceiling; it is a property of the tier. If you send a 1080p request to bytedance/seedance-2.0-fast or -mini, you get a rejection, not a downscale. So the flagship is not “the best tier,” it is “the only tier for 1080p and 4K.” That reframing matters, because the flagship’s price is the entry cost for resolution the cheaper tiers simply cannot produce: $0.34/s at 1080p and $1.37/s at 4K, well above the $0.08–$0.13/s you pay for 720p on Mini or Fast. You are not buying quality, you are buying pixels the other two tiers cannot emit.

The 4-second floor is the same across all three. Every Seedance 2.0 tier refuses clips shorter than 4 seconds. If your format is 2-second loops or 3-second stingers, no tier here can produce them, and the fix is not a different Seedance tier, it is a different model. Wan starts at a 2-second minimum, which is one of the few places it beats Seedance outright; the Seedance 2.0 vs Wan comparison walks through that trade.

Audio is on across all three. Every tier generates a synced audio bed, so a prompt that mentions “footsteps on gravel” gets a matching track and a prompt that says nothing about sound gets whatever the model infers. That is a capability, not a cost lever, and it does not vary by tier, so it never enters the tier decision. It does affect your prompt: name the audio you want on every tier, or you inherit a guess.

The practical rule from this table: pick your tier by resolution ceiling first, budget second, and never by the tier name. “Mini” is a 720p tier at the lowest rate, full stop.

Pricing Math: Same Clip, Three Tiers

Per-second rates are abstract until you multiply them by a clip length. Seedance bills per output second, but the rate depends on the resolution you request — each tier’s headline price is only its 480p floor. So cost equals the tier’s rate at your resolution times duration. Here is the full rate card, then the same clip priced at the resolution each tier is actually used at.

The gaps are small per clip and large per month. On a single 5-second 720p clip, Mini ($0.40) saves $0.25 against Fast ($0.65). That feels like rounding error. A 1080p flagship master of the same length is $1.70 — more than four times a 720p Mini clip — and a 4K plate is $6.85. Multiply any of these by volume and they stop being rounding error.

Same 3,000 clips a month at 5 seconds each. Mini at 720p runs $1,200, Fast at 720p $1,950, the flagship at 1080p $5,100 — a $3,900/month spread driven by both the rate card and the resolution each tier can emit.

Take a small content team generating 3,000 clips per month, 5 seconds each, at 720p. On Mini that bill is 3,000 × 5 × $0.08 = $1,200. On Fast, at the same 720p, it is 3,000 × 5 × $0.13 = $1,950 — a $750/month step for the Fast premium alone. If those clips instead need a 1080p master, only the flagship qualifies, at 3,000 × 5 × $0.34 = $5,100. The gap between a 720p Mini pipeline and a 1080p flagship pipeline is $3,900 a month, or $46,800 a year, part rate card and part resolution. If every clip ships to a feed that caps at 720p, that extra spend buys resolution the platform throws away.

The smarter play is not one tier, it is a ladder. Because all three tiers share the bytedance/seedance-2.0 prefix and one endpoint, routing by tier is a switch statement in your own code, not three integrations.

Generate drafts on Mini at 720p, promote the approved concept to Fast, reserve the flagship for the 1080p or 4K master. A team that drafts 3,000 clips but only finishes 300 pays roughly $1,710/month instead of $5,100 flat on the flagship — 66% less.

Run the numbers on that ladder. A team that generates 3,000 drafts on Mini at 720p but only masters 300 of them on the flagship at 1080p pays 3,000 × 5 × $0.08 + 300 × 5 × $0.34 = $1,200 + $510 = $1,710 a month, two-thirds less than putting every clip through the flagship at 1080p ($5,100). The tier structure is worth far more as a routing tool than as a single choice — and the bigger the resolution jump, the more the ladder saves.

There is a second reason the ladder beats a flat flagship pipeline, and it is mostly about the cost of iteration. In a draft loop you throw away four of every five generations, so running those throwaways at $0.08/s on Mini instead of $0.34/s on the flagship is what makes heavy iteration affordable in the first place. The tier names ByteDance markets as “Fast” and “Mini,” and their market positioning, imply the cheaper tiers also render faster, but ofox does not document per-tier render times, so treat any speed advantage as a hypothesis to test rather than a fact. What is certain is the cost: you only want to pay the flagship’s 1080p rate, and wait its render time, once you have a concept worth mastering. Treat the cheap tiers as your sketchpad and the flagship as your final print.

The one place the ladder does not help is a workflow that needs 1080p on every clip, including drafts. If your review process demands full-resolution previews, Mini and Fast are off the table for that stage because they cannot emit 1080p, and you are back to a flat flagship bill. That is a genuine constraint, not a pricing quirk, so if 720p previews are unacceptable to your reviewers, budget for the flagship’s $0.34/s at 1080p across the board and skip the ladder.

The Fast vs Mini Question the Spec Sheet Can’t Answer

Flagship versus the other two is a clean decision: if you need 1080p or 4K, you take the flagship, because Fast and Mini cannot produce those pixels. The genuinely hard call is Fast versus Mini, because they share the same 720p ceiling and the catalog does not tell you what Fast’s premium buys. At 720p, Mini is $0.08/s and Fast is $0.13/s — a $0.05/s gap that makes Fast 62.5% more expensive at an identical resolution.

Here is the honest state of it. On the datasheet, Fast and Mini are identical except for price: same 720p cap, same modes, same 4-second floor, same audio. Nothing published says Fast produces cleaner motion, better prompt adherence, or fewer artifacts than Mini at matched resolution. Third-party write-ups describe Mini as the faster, cheaper tier and Fast as a step up, but “step up” without a controlled test is a marketing claim, and this guide does not repeat unverified quality rankings as fact.

So the decision procedure is a test, not a spec read:

1. Pick one prompt that represents your hardest real case, not an easy one. Fast camera motion plus a rigid-body action plus a busy background is where cheap video tends to break; a static talking head is not.
2. Generate it on both bytedance/seedance-2.0-fast and bytedance/seedance-2.0-mini at your real duration and 720p, audio on.
3. Judge three things on the output, not the datasheet: does the subject stay coherent through motion, does the prompt’s action actually happen, and does the audio bed sync.
4. If Mini holds up on your prompt, take it and keep the $0.05/s. If Fast is visibly steadier on the motion that matters to you, the premium is cheap insurance on final output. Longer clips compound per-frame drift, so test at 15 seconds if that is your format, not a safe 5-second sample.

That is the whole method. Two API calls settle a question the spec sheet leaves open, and at 720p they cost about $0.65 + $0.40 = $1.05 at 5 seconds. Spending about a dollar to decide a tier that will run thousands of clips is the cheapest research you will do this quarter.

A word on why this matters more than it sounds. The tier names prime you to assume a quality staircase, and if you take that assumption on faith you will default to Fast “to be safe” and quietly pay 62.5% more than Mini would cost, on a resolution ceiling that is identical. Over 3,000 clips a month at 5 seconds that reflexive upgrade is ($0.13 − $0.08) × 5 × 3,000 = $750 you spent on a difference you never tested. The discipline here is simple: do not pay for a quality delta you have not seen with your own eyes on your own prompt. If the test shows Fast is steadier on your hardest motion, buy it with confidence. If Mini holds, keep the money. Either way you are deciding on evidence, not on a marketing label, which is the only defensible way to standardize a tier your pipeline will run for a year.

The Async Workflow Is Identical Across Tiers

Whichever tier you pick, the call is the same. Video generation is not a blocking request the way a chat completion is. The ofox POST /v1/videos call returns immediately with 202 Accepted and a polling_url. The clip renders in the background, and you find out it is ready by polling the task or registering a webhook. None of this changes between Mini, Fast, and flagship, which is exactly why tier routing is cheap: the operational surface is fixed, only the model string moves.

Submit once, then poll GET /v1/videos/{id} or register a webhook until the task reaches a terminal state. The lifecycle is identical whether you called Mini, Fast, or the flagship.

Two operational details save you a support ticket. Poll every 1–2 seconds rather than hammering the endpoint in a tight loop. And for production, prefer a webhook: pass a callback_url at creation and ofox posts a signed payload when the clip finishes. The URL must be public HTTPS; a private or loopback address is rejected at creation time. Because a result URL expires on a TTL, download the clip when it completes rather than assuming the link stays live.

Common Tier-Specific Errors

Most tier failures come from sending a request the tier cannot satisfy, not from the model. The status codes below are observed behavior; ofox’s public docs do not spell out every resolution-mismatch response, so confirm the exact code on your first call. These are the ones that bite when you switch tiers.

The resolution rejections are the common trap when you build a tier ladder. Code that hard-codes "resolution": "1080p" and then routes a job to Mini for cost reasons will fail at submission, because Mini cannot honor 1080p. The clean pattern is to derive resolution from the tier, not to set it independently, so a Mini job requests 720p and a flagship job can request 1080p or 4K.

When to Skip Seedance 2.0 Tiers Entirely

Three tiers cover a lot, but not everything. Reach outside this family when:

• Your clips run under 4 seconds. All three Seedance 2.0 tiers refuse sub-4s output. For 2-second loops and stingers, Wan’s 2-second floor is the deciding capability, walked through in the Seedance 2.0 vs Wan comparison.
• You need Western-market models like Sora or Veo, or Kling specifically. Those run on their own native APIs, not /v1/videos. The Sora 2 Pro vs Veo 3.1 vs Kling 2.6 Pro comparison covers the trade of three separate SDKs versus one shared endpoint, and the Kling 2.6 Pro Video API guide covers Kling end to end.
• You are budgeting across text and image models too, not just video. The full rate-card picture across every model family is in the AI API pricing comparison, which puts video seconds next to text and image tokens in one table.

Match the model to the format first, then optimize the tier. A cheaper tier of the wrong model is still the wrong model.

Try All Three Tiers on One ofox Call

This is where the tier choice stops being a spreadsheet and becomes two API calls. Because all three tiers live behind ofox’s POST /v1/videos, you switch tier by editing the model string. The call is async: submit, get a polling_url, poll until completed.

Run all three tiers on one ofox key. A/B Mini, Fast, and flagship and pick the tier from your own output, not a rate card. Pay-as-you-go, billed per output second by resolution — Mini is $0.08/s at 720p, and the flagship reaches 1080p at $0.34/s.

Python: A/B all three tiers in one loop

import os, time, requests

OFOX = "https://api.ofox.ai/v1"
HEAD = {"Authorization": f"Bearer {os.environ['OFOX_API_KEY']}"}

# 720p is the common ceiling, so all three tiers can run the same request
def generate(model, prompt, resolution="720p"):
    r = requests.post(f"{OFOX}/videos", headers=HEAD, json={
        "model": model,
        "prompt": prompt,
        "duration": 8,
        "resolution": resolution,
        "aspect_ratio": "16:9",
    })
    r.raise_for_status()
    poll = r.json()["polling_url"]        # 202 + polling_url
    while True:
        s = requests.get(poll, headers=HEAD).json()
        if s["status"] in ("completed", "failed", "cancelled", "expired"):
            return model, s
        time.sleep(2)                     # poll every 1-2s, not in a tight loop

prompt = "A skateboarder does a kickflip down a set of stairs, fast camera pan."
for model in ("bytedance/seedance-2.0-mini",
              "bytedance/seedance-2.0-fast",
              "bytedance/seedance-2.0"):
    print(*generate(model, prompt))       # swap the string, same call

Node: the same shape

const OFOX = "https://api.ofox.ai/v1";
const HEAD = { Authorization: `Bearer ${process.env.OFOX_API_KEY}`,
               "Content-Type": "application/json" };

async function generate(model, prompt, resolution = "720p") {
  const res = await fetch(`${OFOX}/videos`, {
    method: "POST", headers: HEAD,
    body: JSON.stringify({ model, prompt, duration: 8,
                           resolution, aspect_ratio: "16:9" }),
  });
  const { polling_url } = await res.json();        // 202 + polling_url
  while (true) {
    const s = await (await fetch(polling_url, { headers: HEAD })).json();
    if (["completed", "failed", "cancelled", "expired"].includes(s.status))
      return { model, s };
    await new Promise(r => setTimeout(r, 2000));    // poll every 1-2s
  }
}

const prompt = "A skateboarder does a kickflip down a set of stairs, fast camera pan.";
for (const model of ["bytedance/seedance-2.0-mini",
                     "bytedance/seedance-2.0-fast",
                     "bytedance/seedance-2.0"])
  console.log(await generate(model, prompt));       // one string swaps the tier

Route by tier in your own pipeline

The routing ladder from the pricing section is a few lines. Draft jobs go to Mini at 720p; approved masters go to the flagship at 1080p or 4K:

def tier_for(stage):
    return {
        "draft":  ("bytedance/seedance-2.0-mini", "720p"),   # $0.08/s at 720p
        "review": ("bytedance/seedance-2.0-fast", "720p"),   # $0.13/s at 720p
        "master": ("bytedance/seedance-2.0",      "1080p"),  # $0.34/s; $1.37/s at 4K
    }[stage]

model, res = tier_for("draft")
model, s = generate(model, "product on a turntable, soft studio light", res)

One auth, one schema, three price points. That is the practical reason to A/B the tiers instead of picking on the datasheet.

FAQ

What is the difference between Seedance 2.0, Fast, and Mini? Resolution ceiling and price, not a documented quality ranking. Flagship bytedance/seedance-2.0 reaches 4K; Fast and Mini cap at 720p. Price scales with resolution — the $0.04 / $0.06 / $0.07 you see are 480p floors. At 720p, Mini is $0.08/s, Fast is $0.13/s, and the flagship is $0.16/s; the flagship then hits $0.34/s at 1080p and $1.37/s at 4K. All three share t2v, i2v, v2v, audio, a 4-second minimum, and one POST /v1/videos endpoint.

Which Seedance 2.0 tier is cheapest? Mini. At 720p it bills $0.08/s, half the flagship’s 720p rate and about 38% below Fast. On a 5-second 720p clip that is $0.40 versus $0.65 on Fast. Mini caps at 720p.

Does Seedance 2.0 Fast or Mini support 1080p or 4K? No. Both cap at 720p. Only the flagship bytedance/seedance-2.0 lists 1080p ($0.34/s) and 4K ($1.37/s).

What is the model ID for each tier on ofox? bytedance/seedance-2.0 (flagship), bytedance/seedance-2.0-fast, and bytedance/seedance-2.0-mini. Same key, same endpoint.

Is Seedance 2.0 Fast worth it over Mini? At their shared 720p ceiling, Fast is $0.13/s versus Mini’s $0.08/s — a $0.05/s premium, about 62.5% more. The case for it is output quality on your prompts, which the spec sheet does not settle. Test both at your real duration first.

How much does a 5-second clip cost on each tier? At the resolution each tier is used at: Mini $0.40 (720p), Fast $0.65 (720p), flagship $1.70 (1080p) or $6.85 (4K). Billing is per second, so a 10-second clip is exactly double.

Should I use Seedance 2.0 or Wan? Different decision. Wan starts at $0.10/s (720p; $0.15/s at 1080p), caps at 1080p, and starts at 2 seconds. See the Seedance 2.0 vs Wan comparison; this guide assumes you already chose Seedance.

Is there a newer Seedance model than 2.0? ByteDance previewed Seedance 2.5 with native 4K and longer single-shot clips in mid-2026. The three 2.0 tiers here are what is live on ofox as of 2026-07-20.

Sources Checked for This Refresh

• ofox model catalog, per-second video pricing shown as “from” floor rates: https://ofox.ai/video
• ofox model detail page, flagship tier (480p–4K per-resolution pricing): https://ofox.ai/models/bytedance/seedance-2.0
• ofox model detail page, Fast tier (480p / 720p pricing): https://ofox.ai/models/bytedance/seedance-2.0-fast
• ofox model detail page, Mini tier (480p / 720p pricing): https://ofox.ai/models/bytedance/seedance-2.0-mini
• ofox model detail page, Wan 2.7 (720p / 1080p pricing): https://ofox.ai/models/alibaba/wan-2.7
• ByteDance Seed, Seedance overview: https://seed.bytedance.com/en/seedance
• Volcano Engine documentation portal: https://www.volcengine.com/docs/82379
• Seedance 2.0 vs Wan comparison (Seedance-vs-Wan decision): /en/blog/seedance-2-0-vs-wan-video-api-2026
• Sora 2 Pro vs Veo 3.1 vs Kling 2.6 Pro comparison: /en/blog/ai-video-generation-apis-sora-veo-kling-compared-2026
• Kling 2.6 Pro Video API guide: /en/blog/kling-2-6-pro-video-api-english-complete-2026
• AI API pricing comparison: /en/blog/ai-api-pricing-comparison-2026