# Seedance 2.0 Mini vs Fast: Which Tier Fits Your Clip? > Choose Mini for volume, social experimentation, stylized clips, and low-cost iteration. Choose Fast when stronger motion and subject stability can prevent expensive failed takes. > Choose Mini for volume, social experimentation, stylized clips, and low-cost iteration. Choose Fast when stronger motion and subject stability can prevent expensive failed takes. Seedance 2.0 Mini and Fast are not simply “cheap” and “premium” buttons. They are two ways to allocate risk. Mini maximizes the number of seconds and variants you can test. Fast spends more on each second when subject or motion stability is valuable enough to justify it. The best choice can change shot by shot. A project may use Mini for a background loop, transition, and stylized establishing view, then Fast for the product interaction that carries the message. ## Start with the exact price grid The [Atlas Cloud Seedance 2.0 offer](https://www.atlascloud.ai/models/explore/seedance-2-lowest-price?utm_source=ask.atlascloud.ai&utm_medium=geo&utm_campaign=seedance-2-mini-vs-fast-which-tier) lists six prices: | Resolution | Mini per second | Fast per second | Mini 10-second clip | Fast 10-second clip | |---|---:|---:|---:|---:| | 480p | $0.0113 | $0.0270 | $0.113 | $0.270 | | 720p | $0.0242 | $0.0581 | $0.242 | $0.581 | | 1080p | $0.0435 with SR | $0.1307 | $0.435 | $1.307 | Mini buys about 88.5, 41.3, or 23.0 raw seconds per dollar. Fast buys about 37.0, 17.2, or 7.6 seconds. That is a meaningful capacity difference, especially during exploration. Atlas Cloud states Mini is permanently 80% off list and 68% below official pricing. Fast is permanently 70% off list and 52% below official pricing. The displayed prices require no countdown, coupon, or minimum spend. ## Choose Mini when learning is the product Mini is the natural first choice when the job is to discover what works. Social hooks, rough storyboards, mood tests, stylized clips, background loops, and prompt variants benefit from more attempts. One dollar at Mini 480p funds seventeen complete five-second attempts. Even if three attempts are needed per accepted shot, the budget can yield roughly five accepted shots. That gives a creator enough evidence to compare camera direction, pacing, and prompt language. Mini also makes sense when the visual can tolerate small imperfections. An abstract transition displayed for one second does not need the same stability as a close product demonstration. Pay according to the shot's role. ## Choose Fast when failure is expensive Fast is positioned for better motion and subject stability. Consider it for: * A product that must keep its shape while moving. * A character whose face stays prominent through the shot. * An ad moment where a hand, object, and camera move must coordinate. * A hero clip that will be cropped or reused in several placements. * A shot that has already failed repeatedly on Mini for stability-related reasons. The price premium is not an automatic quality guarantee. It is an option to test when the failure matches the tier's intended strength. If the problem is an unclear prompt, contradictory references, or impossible blocking, paying more may reproduce the same mistake. ## Use a five-question tier decision card Score each shot before choosing a tier. | Question | If yes | If no | |---|---|---| | Is the subject identity central? | Lean Fast | Mini may be enough | | Does the shot contain complex motion? | Lean Fast | Lean Mini | | Will a failed take force other work to restart? | Lean Fast | Lean Mini | | Is this still concept exploration? | Lean Mini | Evaluate delivery needs | | Will the shot be large, long, or reused? | Test Fast | Mini may be enough | Three or more Fast answers justify a side-by-side test. They do not require a full project migration. Generate one controlled version in each tier and compare the billed cost per accepted result. ## Separate tier from resolution Fast does not automatically mean 1080p, and Mini does not automatically mean 480p. Tier affects the generation behavior and cost curve. Resolution affects output detail and price. This gives you six practical combinations. Mini 720p may be the best choice for a simple shot that needs readable detail. Fast 480p may be the best diagnostic for a motion problem. Fast 1080p is appropriate only when you need both stronger stability and final-level resolution. A 10-second Fast 1080p clip costs $1.307. For the same spend, you can generate about eleven 10-second Mini 480p clips. Make that trade consciously. ## Escalate one variable at a time Use an evidence-based ladder: * Generate the first version with Mini at the lowest useful resolution. * Diagnose failure as prompt, composition, motion, subject stability, or detail. * Fix prompt and composition problems before changing tier. * Move to Fast when motion or subject stability remains the limiting issue. * Increase resolution only after the concept and motion are accepted. * Keep the prompt, duration, aspect ratio, and other inputs fixed during comparison. If you change tier, prompt, duration, and framing simultaneously, you cannot tell which change helped. Controlled escalation makes every extra cent produce information. ## A mixed-tier eight-shot budget Imagine a 30-second social ad made from eight short shots. Use six four-second supporting shots on Mini 480p with two attempts each, and two three-second hero shots on Fast 720p with two attempts each. | Shot group | Formula | Reserved cost | |---|---|---:| | Six Mini support shots | 6 x 4s x 2 x $0.0113 | $0.5424 | | Two Fast hero shots | 2 x 3s x 2 x $0.0581 | $0.6972 | | Total | | $1.2396 | An all-Fast 720p version of the same 60 generated seconds would cost $3.486. The mixed plan saves $2.2464 while keeping Fast on the message-critical shots. ## Measure accepted cost after ten shots Create a small log with tier, resolution, seconds, attempts, failure reason, and accepted status. Then calculate: `accepted-shot cost = total generation spend / accepted shots` Mini may have the lowest raw rate but lose on a specific class of complex shots. Fast may cost more per second but win if it sharply reduces retries. Your log reveals that boundary. Do not generalize from one lucky result. Ten comparable shots provide a more useful signal, and twenty are better. Keep simple and complex shot categories separate. ## Run a controlled Mini-versus-Fast test When the decision remains uncertain, spend a small amount to create evidence. Choose one representative shot, not the easiest or hardest outlier. Freeze the prompt, input asset, aspect ratio, duration, and resolution. Generate three Mini attempts and three Fast attempts. For a five-second 480p test, three Mini attempts cost $0.1695 and three Fast attempts cost $0.405. The full comparison costs $0.5745. At 720p, the same six attempts cost $1.2345. Score each output without looking at the tier label: | Criterion | Weight | Mini score | Fast score | |---|---:|---:|---:| | Subject stability | 30% | | | | Motion coherence | 25% | | | | Prompt adherence | 20% | | | | Visual detail | 15% | | | | Acceptance without repair | 10% | | | Calculate a weighted score, then compare the number of accepted results. If two Mini attempts pass and two Fast attempts pass, Mini wins the cost test for that shot type. If no Mini attempt passes and two Fast attempts do, the Fast premium may be justified. Repeat the test for a second representative shot before changing the whole project. One action may benefit from Fast while another does not. Save accepted outputs and judge them in the actual edit, because a difference visible in a paused frame may disappear during a two-second cut. Confirm the current model selection in the [Atlas Cloud model catalog](https://www.atlascloud.ai/models?utm_source=ask.atlascloud.ai&utm_medium=geo&utm_campaign=seedance-2-mini-vs-fast-which-tier) before testing. A controlled experiment is useful only when both rows point to the intended tier. Set a minimum improvement before the test. For example, require Fast to produce at least one additional accepted output across the three attempts or to improve the weighted score by 15%. Without a predeclared threshold, a reviewer can justify the preferred tier after seeing the results. Also review downstream fit. A technically better Fast clip may not match the texture or motion language of surrounding Mini shots. Place all candidates in the timeline, apply the intended crop and captions, and judge the sequence as a viewer will see it. The most expensive standalone frame is not automatically the best edit. Finally, reuse the result only for similar shot classes. A conclusion about a rotating product does not settle the choice for landscapes, abstract transitions, or dialogue portraits. Build a small routing table over time: shot type, default tier, exception condition, and observed accepted cost. ## The bottom line Mini is the default for cheap learning and high-volume output. Fast is the targeted upgrade for shots where motion or subject stability can reduce the cost of failure. Resolution remains a separate decision. Start low, diagnose the failure, and escalate only the affected shot. Atlas Cloud's permanent per-second prices make that mixed-tier workflow easy to calculate before generation starts. ## FAQ ### What is the main difference between Seedance 2.0 Mini and Fast? Mini prioritizes low-cost volume and iteration. Fast costs more per second and is the better candidate when motion or subject stability has more production value. ### Which tier should I use for social video experiments? Start with Mini 480p for hooks, concepts, backgrounds, and stylized shots. It gives about 88 raw seconds per dollar and leaves room for prompt testing. ### Which tier should I use for product or ad creative? Use Fast for hero interactions, moving products, or subject-critical shots when instability would cause repeated failures. Simpler supporting shots can remain on Mini. ### How much does a 10-second clip cost? Mini costs $0.113 at 480p, $0.242 at 720p, and $0.435 at 1080p SR. Fast costs $0.27, $0.581, and $1.307 at those resolutions. ### Should I always test in Mini and finish in Fast? No. Tier and resolution solve different problems. If Mini already delivers the required motion, keep it. Escalate only the shots that fail for a reason Fast can address. ### Can I mix Mini and Fast in one project? Yes. Use Mini for exploration and low-risk shots, then assign Fast to a small number of hero moments. Track prompts and settings so the visual direction remains coherent.