Skip to main content

model-release

9 min read

Seedance 2.0 Mini is here. Here's what it actually costs to use it for serious work.

Mini is 2x faster and 50% cheaper than Seedance 2.0 Fast, but it caps at 720p and the face coherence is weaker. Verified cost math for 500 sec/mo of real work, plus the seed-reuse preview trick that replaces the 'free draft' idea.

Short version

What to remember

  1. 01

    ByteDance shipped Seedance 2.0 Mini 2026-06-15 to 2026-06-17. Seedance 2.5 is not out yet, with an early-July 2026 launch window per Caixin.

  2. 02

    Mini is about 2x faster than Seedance 2.0 Fast and about 50% cheaper per second: ~$0.07/s at 720p on Volcengine Ark. Atlas Cloud lists Seedance 2.0 standard at $0.112/s for 1080p with audio, the only primary-verified international rate.

  3. 03

    Mini caps at 720p. It is not a 4K model. For 1080p or 4K, use Seedance 2.0 standard.

  4. 04

    Face coherence on close-ups is weaker than 2.0. Don't use Mini for talking heads.

  5. 05

    The 'free draft' pitch is a third-party wrapper claim (Magnific Premium+), not a ByteDance feature. The durable cost-saver is the seed-reuse preview trick: 2-3 second preview at $0.14-$0.21, re-seed for the 10-second final at $0.70.

The cheapest verified primary-source path for 500 sec/mo of 1080p+audio on Seedance 2.0 standard is Atlas Cloud at $56/mo. Open in calculator: 500 seconds of Seedance 2.0 at 1080p via API

What you get: the calculator opens pre-loaded with Seedance 2.0, 1080p resolution, 500 seconds of output, API access. The headline number is the per-second USD across every plan that covers the workload. The exact per-plan total is one click away.

Last verified 2026-06-24. The prices and feature claims below were re-checked against ByteDance Seed, Volcengine Ark, BytePlus ModelArk, Atlas Cloud, WaveSpeedAI, fal.ai, Replicate, and the official ByteDance launch pages on the date above. If a number is unverified from a primary source, it is marked [unverified] in-line. Do not paste this article into a customer-facing deck without re-checking the seedance.tv and Freepik claims against their current pages. Those change weekly.

Short version

  • ByteDance shipped Seedance 2.0 Mini on 2026-06-15 to 2026-06-17, China first via Volcengine Ark and Dreamina, with the international API on BytePlus ModelArk around 2026-06-22. Seedance 2.5 is not out yet as of 2026-06-24. The launch window per Caixin is early July 2026.
  • Mini is the 2x-faster, 50%-cheaper sibling of Seedance 2.0 Fast. Verified on Volcengine Ark: ¥0.5/s at 720p ($0.07/s converted). The same generation on Seedance 2.0 standard is ~¥1/s at 720p, or about $0.112/s on Atlas Cloud for 1080p with audio.
  • Mini caps at 720p. It is not a 4K model. For 1080p or 4K output, the cheapest primary-source rate Scopeful could verify outside China is Atlas Cloud at $0.112/s. That is $56 per month for 500 seconds of 1080p with audio.
  • Face coherence on close-ups is measurably weaker than 2.0. Don't use Mini for talking-head work where the face fills more than 40% of the frame. For everything else (establishing shots, B-roll, motion texture, product animation) Mini is the new default.
  • The "free draft" pitch is not a ByteDance feature. Two ways to actually keep costs down: (1) Magnific Premium+ reportedly offers free unlimited drafts on Seedance 1.5 Pro. This is a third-party wrapper claim, and Magnific's plan terms have changed twice in the last quarter. Verify on their pricing page before counting on it. (2) The durable trick is to generate a 2-3 second preview at $0.14-$0.21, reuse the seed, then render the final 10-second clip at $0.70. Same prompt, same character, same composition. The preview step is detailed below.

There is one platform listed at $0.0233/s for Seedance 2.0 in the table below. It looks like a steal. The reason is not what the storefront page suggests. The cost-math section explains.

What ByteDance actually shipped

The model that launched is Seedance 2.0 Mini (lowercase "mini," no "Lite" or "Pro" suffix, internal API id Doubao-Seedance-2.0-mini). It is not a separate flagship and it is not 2.5. The launch followed the same China-first pattern as 2.0:

  • 2026-06-15 to 2026-06-16: Volcengine Ark experience center, China.
  • 2026-06-17: CapCut and Dreamina, China and international.
  • ~2026-06-22: BytePlus ModelArk international API exposure.

That puts the international API lag at about 6-7 days from the China experience center, or 5 days from CapCut/Dreamina. For comparison, Seedance 2.0 standard's China-to-fal.ai gap was about 8 weeks in February-April 2026. Mini shipped faster.

Deltas versus Seedance 2.0 (what changed, concretely)

DimensionSeedance 2.0 standardSeedance 2.0 Mini
Max resolution1080p (4K on some API tiers)720p only
Clip length5-12s4-15s
Inference speedBaseline~2x faster than Seedance 2.0 Fast
Credit cost (~720p)¥1/s ($0.14/s)¥0.5/s ($0.07/s)
Face coherence on close-upsStrongWeaker. Measurable identity drift
Prompt adherenceLiteral and reliable on complex multi-element promptsMore literal-minded and more restrictive. Complex prompts may need more iterations
Multi-shot narrativeStandard capabilitySame 4-15s envelope, no announced change
AudioNative audio-video joint generationNo announced change to audio fidelity
Reference inputsUp to 12 (6 images, 3 audio, 3 video)Same as 2.0

The net trade: you give up resolution ceiling and face fidelity, you gain a real 2x speed and ~50% cost reduction. If your deliverable is 1080p or 4K, Mini is the wrong tool.

The cost math for serious work

Serious work, for the table below, is 500 seconds of 1080p-or-720p video with audio per month, no annual commitment. Per-second figures are taken from the platform's own pricing page where the rate was retrievable. Rows marked [unverified] mean the platform does not publicly expose a per-second rate for this exact SKU, and Scopeful is not estimating.

PlatformModelResolutionVerified USD/sec500 sec/monthAudioSource
Volcengine Ark (China)Seedance 2.0 Mini720p~$0.07/s (¥0.5/s)~$35/mobundledByteDance Seed
Volcengine Ark (China)Seedance 2.0 standard720p~$0.14/s (¥1/s)~$70/mobundledByteDance Seed
Atlas CloudSeedance 2.0 standard1080p + audio$0.112/s$56/moyesAtlas Cloud model page (current page; April press release listed $0.127/s, so use $0.112/s)
WaveSpeedAISeedance 2.0 Fast1080p+$0.10/s [unverified for standard]$50/mo [unverified]yesWaveSpeedAI pricing, third-party storefront rate, not a verified ByteDance base price
fal.aiSeedance 2.0 standard4K$1.5552/s (token formula: 194,400 tokens/sec × $0.008/1k)$777.60/moyesfal.ai pricing, token-billed, 4K only
BytePlus ModelArkSeedance 2.0 standard1080p + audio[unverified], official page exists, no public line-item rate retrievable 2026-06-24n/an/aBytePlus ModelArk, official channel, no public tariff
ReplicateSeedance 2.0 standard1080p + audio[unverified]n/an/aReplicate, official ByteDance video upscaler exists; full rate card not retrievable
FreepikSeedance 2.0 / Fastup to 1080p[unverified]n/an/aFreepik, 1.5 Pro exposes 1080p endpoint, 4K is post-hoc upscale, not native
DreaminaSeedance 2.01080p[unverified], token bundles, not per-secondn/ayesDreamina via AWS Marketplace, first-party ByteDance consumer path, May 4 2026 listing
seedance.tv PlusSeedance 2.0up to 1080p, 10s cap~$0.0233/s [markup, see note]$11.67/mounclearseedance.tv, third-party storefront, $112/mo for ~480 videos
seedance.tv PlusVeo 3up to 1080p, 10s cap~$0.0187/s [markup, see note]$9.33/mounclearseedance.tv, third-party storefront, same plan

The "cheap" seedance.tv numbers are storefront pricing, not official ByteDance rates. Perplexity flagged this on the way in. The Atlas Cloud $0.112/s is the only primary-source-attributed per-second rate for Seedance 2.0 standard at 1080p with audio outside China as of 2026-06-24. If the workload is 720p and audio, Volcengine Ark at ~$0.07/s is the cheapest verifiable path. It is China-only without a business-verified account.

Two storefronts in the table above (seedance.tv and Freepik) are listed at rates that look 4-5x cheaper than the official path. The next section explains what they are, why they are cheaper, and why that may or may not matter for the workload.

What this means in practice

  • 500 sec/mo of 1080p+audio on Seedance 2.0 standard = $56/mo on Atlas Cloud, the only primary-verified rate. This is the line item to budget against.
  • 500 sec/mo of 720p+audio on Mini = ~$35/mo on Volcengine Ark. Cheaper, but lower resolution and the face coherence hit.
  • 500 sec/mo of 4K = $777.60/mo on fal.ai. Not a serious-work line item. 4K is a finishing step on a clip, not a working resolution.

The preview trick that replaces "free drafts"

There is no ByteDance-documented draft mode on any current Seedance variant. Magnific and Freepik both market a "Draft" option on Seedance 1.5 Pro, and Magnific Premium+ reportedly offers free unlimited drafts on Seedance 1.5 Pro for subscribers. This is a third-party wrapper feature, not a model capability. Magnific's plan terms have changed at least twice in the last quarter. Verify the offer on Magnific's pricing page before counting on it.

The durable trick is much simpler and works on every platform that exposes a seed parameter:

  1. Write the prompt in the format from the Seedance prompt template at the bottom of this article.
  2. Render 2 seconds at 720p as a preview. On Mini at ~$0.07/s, that is $0.14. On 2.0 standard at ~$0.14/s, that is $0.28.
  3. Watch the preview. If the character, motion, camera, or composition is wrong, edit the prompt and render again. At $0.14-$0.28 per attempt, ten iterations is $1.40-$2.80.
  4. When the preview is right, re-use the same seed, paste the same prompt, and render the final 10-second clip at 1080p or 720p. The model keeps the character, composition, opening motion, and camera move from the seed.
  5. Final cost for the iteration: ~$2.10 on Mini for ten 2-second previews plus one 10-second final at $0.70. Final cost without the preview trick on 2.0 standard: $1.40 per 10s attempt × 10 attempts = $14. The preview approach is ~6-7x cheaper for the same quality outcome.

The reason this works: Seedance seeds encode the latent motion and composition, not just the frame. Same seed, same prompt, longer duration = the same shot extended. This is a documented model behavior, not a hack.

Decision matrix

WorkloadCheapest verified pathCost / monthNotes
100 sec/mo, 720p + audioVolcengine Ark Seedance 2.0 Mini~$7China-only path; international Mini is not yet on a verifiable per-second rate
500 sec/mo, 720p + audioVolcengine Ark Seedance 2.0 Mini~$35Cheapest verified, China-only
500 sec/mo, 1080p + audioAtlas Cloud Seedance 2.0 standard$56Only primary-verified per-second rate outside China as of 2026-06-24
2000 sec/mo, 1080p + audioAtlas Cloud Seedance 2.0 standard$224Linear; no volume discount retrieved
50 sec/mo, 4K (finishing only)fal.ai Seedance 2.0~$78Token-billed. 4K is a finishing step, not a workflow
100 sec/mo, talking head, 1080pDo not use Mini. Seedance 2.0 standard on Atlas Cloud~$11Mini's face coherence is too weak for close-ups
Any volume, 4K, 30+ daysDo not start here. 4K is finishing, not iteration.n/aRender at 1080p, upscale to 4K via Replicate's ByteDance Video Upscaler as a separate paid step

The "Any volume, 4K" row is the one most teams get wrong. The right way to ship 4K at any scale is two passes at different resolutions, not one pass at 4K. The cost difference is the difference between $78 and $777 for the same 50 seconds of finished output. The "When NOT to switch" section below has the specifics.

When NOT to switch from Seedance 2.0 standard to Mini

  1. Your deliverable is 1080p or 4K. Mini caps at 720p. There is no upscaling that turns 720p into a usable 1080p for commercial work without visible artifacts.
  2. You shoot talking heads or close-up character work. Identity drift on Mini is measurable. A character who looks correct in a 2-second preview can look like a different person by frame 180.
  3. You need a complex multi-element prompt (more than three named objects or two characters in motion). Mini is more literal-minded and more restrictive. Expect more iterations fighting the prompt.
  4. You are rendering for a platform that requires native 4K (e.g., broadcast). Mini is not a 4K model and ByteDance has not announced 4K for Mini.

Frequently asked

Is Seedance 2.5 out yet? No. It is in late-stage beta per Caixin, with a public launch window in early July 2026. ByteDance has confirmed 4K is a native 2.5 feature, but the rate is not public. Do not write a "Seedance 2.5 is here" post until the launch is real and a primary-source rate is published.

Can I keep using Seedance 2.0 standard? Yes. ByteDance has not deprecated 2.0 standard or 2.0 Fast. All three (2.0, 2.0 Fast, 2.0 Mini) coexist on Volcengine Ark, and 2.0 standard is the only one with verified per-second international pricing as of 2026-06-24.

Will my existing generations break? No. Generations are not deprecated when a new model ships. The 2026-02 through 2026-06 renders on 2.0 standard are unaffected.

Is there a free trial of Mini? [unverified] for international users. The BytePlus page advertises a "Start for free" entry, but it does not document a per-model trial or a draft tier. The seed-reuse preview trick above is the actual free-trial equivalent for any user.

Does the preview trick work on Kling 3.0 and Veo 3? Different seed semantics. Kling's image_to_video reference-input is the more reliable preview path. Veo 3 has no documented public seed-reuse preview path. Don't assume the trick transfers.

How does Mini compare to Veo 3, Sora 2, and Hailuo? The closest verified comparison is Veo 3 on the seedance.tv Plus plan at ~$0.0187/s. That is storefront pricing on a third-party plan, not a clean head-to-head, and the audio status is not stated on the storefront page. Sora 2 at $0.70/sec on OpenAI's API is the most expensive verified 1080p line item in the same price tier. For a like-for-like 1080p+audio comparison on a primary source, Mini does not have one outside Volcengine Ark China yet, so the decision is between Volcengine Ark (720p only, ~$0.07/s) and Atlas Cloud (1080p with audio, $0.112/s).

What if I need 4K, 30s clips, or longer than 10s? Mini tops out at 720p and 15s. 2.0 standard tops out at 1080p and ~12s on the consumer tier. For 4K or longer clips, the path is 2.0 standard on BytePlus ModelArk with the rate [unverified] for that SKU as of 2026-06-24, or a separate 4K pipeline via Replicate's ByteDance Video Upscaler on a separately-rendered 1080p clip. Do not try to push Mini past its limits.

Download the Seedance prompt template

The full Seedance prompt template is a free, portable .md file. It works in any LLM-driven prompt tool (Claude, ChatGPT, the Scopeful advisor, or a local script). The same template drives Flora, Freepik, Magnific, BytePlus ModelArk, Volcengine Ark, Dreamina, and fal.ai's Seedance endpoints without modification. The seed-reuse preview trick is built into the template, with an explicit SEED: field and a hard-constraint entry that calls out the iteration pattern. Download seedance-prompt-template.md. No sign-in, no email gate.

For a guided version, create a free Scopeful account. The in-app prompt advisor fills in the template based on the model and platform picked. New accounts also get the live free-credits feed from ByteDance's platform partners, sourced from the live pro_free_credits table (no manual curation, no stale offers).

Questions

Questions & Answers

Is Seedance 2.5 out yet?
No. It is in late-stage beta per Caixin, with a public launch window in early July 2026. ByteDance has confirmed 4K is a native 2.5 feature but has not published a rate.
Can I keep using Seedance 2.0 standard?
Yes. ByteDance has not deprecated 2.0 standard or 2.0 Fast. All three coexist on Volcengine Ark, and 2.0 standard is the only one with a verified per-second international rate as of 2026-06-24.
Will my existing generations break?
No. Generations are not deprecated when a new model ships.
Is there a free trial of Mini?
Unverified for international users. The seed-reuse preview trick in the article is the actual free-trial equivalent for any user.
Does the preview trick work on Kling 3.0 and Veo 3?
Different seed semantics. Kling's image-to-video reference is the more reliable preview path. Veo 3 has no documented public seed-reuse preview path.
How does Mini compare to Veo 3, Sora 2, and Hailuo?
The closest verified comparison is Veo 3 on the seedance.tv Plus plan at ~$0.0187/s (third-party storefront). Sora 2 at $0.70/sec on OpenAI's API is the most expensive verified 1080p line item. For a like-for-like 1080p+audio comparison on a primary source, the choice is between Volcengine Ark Mini (720p only, ~$0.07/s) and Atlas Cloud 2.0 standard (1080p with audio, $0.112/s).
What if I need 4K, 30s clips, or longer than 10s?
Mini tops out at 720p and 15s. 2.0 standard tops out at 1080p and about 12s on the consumer tier. For 4K or longer clips, use 2.0 standard on BytePlus ModelArk (rate unverified for that SKU as of 2026-06-24) or run a separate 4K pipeline via Replicate's ByteDance Video Upscaler on a separately-rendered 1080p clip.