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.

Seedance 2.5 Review: Uncensored 30-Second Takes, High Price

Pick Seedance 2.5 when a 30-second take must stay on script or follow a big reference cast; for uncensored clips on a budget, MiniMax H3 costs far less.

Spicy Index

82.2/ 100

Freedom Score

83.6/ 100

Range if the rest pass: 83.6–98.9

  • Mild

  • Moderate

  • Strong

  • Very strong

  • Maximum

Price
$1.39 per 5-second clip
Median time
4.5 min
Success rate
Collecting data

Seedance 2.5 passed all 5 content rungs we ran on 24 Sep 2026 across text, a still and references, from L1 Suggestive (lingerie, a slipping strap) through L2 Mature themes, L3 Strong and L4 Intense (an intimate scene between two adults) to L5 Extreme, in two kinds: L5a (consensual restraint and impact play between adults) and L5b (cinematic gore). L4 Intense passed 8 of 9 runs, and every L1 to L3 run passed. Pick it for a 30-second take with sound that stays on script, at $0.2773 per second at 720p.

The verdict

On 14 September we typed our Maximum-level prompt into Seedance 2.5 text-to-video. The clip came back as written: every element we named on screen, nothing toned down. The next day the same scene went through from a still and from reference images. On 24 September we went back and ran every level from Mild to Very strong three times on each task: 32 of 33 clips came back as written, and the one exception kept the scene implied rather than delivering it as asked.

Pick Seedance 2.5 when one shot has to run long and stay on script. It does up to 30 seconds in a single take, with sound, from text, a still or a stack of references, and it holds a character through the whole thing. Wan 3.0 also runs 30 seconds with sound for about a third of the price and rendered every rung we ran, but it pushes scenes past what you wrote. Skip Seedance 2.5 for cheap drafts: MiniMax H3 rendered every ladder clip we sent for a fraction of the price, in takes of up to 15 seconds.

How it scores. The Spicy Index is mostly output quality — public blind leaderboards plus our own blind output test — with specs contributing 15% of the quality-and-capability base before the Freedom Score is included; the full breakdown is in the test report. On specs, it offers sound, reference inputs and first-and-last-frame control, natively up to 720p (the Index counts 1080p here as upscaled, not native), and it has no video-to-video task or LoRA support. The Freedom Score is explained under How far it goes.

Try text-to-video → · Try image-to-video → · Try reference-to-video →

Key facts

  • Freedom Score: on this page's score card, recalculated as runs are added; all 5 rungs passed on 24 Sep 2026 across text, a still and references.
  • Pass rate: 32 of 33 runs on 24 Sep 2026 came back as written; the one miss implied the scene instead of showing it.
  • Spicy Index: see the score card above.
  • Price: $0.1234 per second at 480p, $0.2773 at 720p and $0.6823 at 1080p, sound included.
  • Limits: 4 to 30 seconds per take, up to 1080p, up to 30 reference images, 10 clips and 10 audio files.
  • Sources: every scored run is in the Seedance 2.5 test report, the scoring rules are on our review methodology page, and the full parameter list is on the Seedance 2.5 model page.

What we liked and what held it back

What we liked

  • One take, up to 30 seconds. Any length from 4 to 30 seconds on all three tasks. Our 15-second last-train example cut itself into four shots and kept the same woman in the same coat through every cut.
  • Uncensored from every starting point. Text, a still or references: every level we ran, from Mild to Maximum, came back as written, with nothing swapped for a safer scene.
  • The biggest reference budget we sell. Up to 30 images, 10 video clips and 10 audio files in one request. Our three-image example kept her wardrobe from two person shots and put her in the room from the third, in one unbroken shot.
  • Sound at no extra cost. Audio is on by default at the same per-second price, and every sound-on example came back with a track: room tone, a whispered line, a saxophone through the wall.
  • Motion that commits. Give it a physical event and it plays it out: the camera charges in and whips around her while the glass shards spiral.
  • A seed and a closing frame. Fix the seed to see exactly what one prompt edit changed; add a closing frame on image-to-video to land the shot on a picture you chose.

What held it back

  • It's expensive. $0.62 for 5 seconds at 480p, $1.39 at 720p, $3.41 at 1080p; a 30-second take at 720p is $8.32. Only 2.5 Spicy costs more per 720p second in our catalogue.
  • "Kiss" gets refused. In our example shoots, prompts with a kiss were turned away by Seedance's own check before a frame rendered, at no charge. A whisper at her ear went straight through.
  • It cuts when you didn't ask. A slow push-in on our night-train prompt came back as a wide shot followed by a close two-shot. Add "one continuous shot, no cuts".
  • "Don't" reads as a suggestion. We wrote "neither looks at the other" and they ended face to face. Write the look you want instead.
  • Image-to-video can leave your still behind. Describe a different room and it shows your still for a split second, then cuts to the one you wrote. Keep the motion inside the still's own scene, and crop the still to your final shape first: there is no aspect-ratio option.
  • It takes minutes. Real jobs take a median of just over four minutes, and 30 seconds at 1080p takes longer.

What the clips look like

The gallery on this page holds published, safe-for-work examples from the Seedance 2.5 model page, shown as rendered. Our ladder test clips aren't shown; we report the results only.

Long takes hold together. The last-train prompt asks for thirty seconds; we ran it for 15 at 1080p. Seedance cut it into four shots on its own (alone on the seat, passengers boarding, the man pressed behind her, a close-up of the two of them) and the same woman in the same cream coat carries through every cut. That consistency is why people pick it for scenes longer than one beat.

Closeness without the word. Our first version of the jazz-bar shot had a kiss in it and was refused before rendering. The rewrite has her whisper against the other woman's ear: lips at the cheek, a wide laugh, foreheads together, and the whispered line on the soundtrack.

Big moves from a still. Given a frame and a physical event, image-to-video commits: the camera charges in, the shards spiral, the rain lets go in one blast.

The ladder clips, as results only. From text, the 14 September Maximum-level clip kept every element of the prompt on screen, with the camera kept wide. From a still, base 2.5 staged a calm tableau: the whole scene held side-on, every prop in place, and the camera never moves. Our prompt asked for a calm mood and an intense one at once, and 2.5 chose calm; Seedance 2.5 Spicy chose intense with the same inputs.

Prompts that worked

The formula: the adult subject and what they wear, two or three beats of action joined by "then", one camera move, the light, the sound. On image-to-video, write only what moves. For the upper levels, keep the same structure, say who is in the shot and who does what, and ask for one mood and one continuous shot. We don't publish our test prompts themselves.

1. Closeness without the word "kiss" (text-to-video, 720p, 5 s). The exact prompt behind the jazz-bar clip.

text
Warm, clearly lit close two-shot. Two adult women stand very close in the amber glow of a jazz bar's back hallway, one in a black velvet slip, the other in an open white shirt. One leans in and whispers something against the other's ear; the other closes her eyes and breaks into a wide laugh, then they rest their foreheads together. Handheld 35mm, warm practical lights, shallow focus, muffled saxophone, a whispered line and soft laughter.

Run this prompt on Seedance 2.5 →

2. A long take that edits itself (text-to-video, 1080p). Set the length on the page; the words "thirty seconds" don't do it. For one unbroken take, add "one continuous shot, no cuts".

text
Cast: one man and one woman. Thirty seconds on a last train at night: she sits alone in an empty carriage with her coat falling open, and at each stop more people board until the two of them are pressed chest to back against the door, her head tipped back on his shoulder, neither of them speaking. Warm carriage light against tunnel black. Cinematic, shallow depth of field, heavy film grain, colour.

Run this prompt on Seedance 2.5 →

3. Your character in a new room (reference-to-video, 1080p, 9:16, 6 s). Three references in upload order: two of the woman, one of the room. Say what each one is before the action.

text
Put the woman from @Image1 and @Image2 into the candlelit bathroom of @Image3. She steps behind the translucent shower curtain so she reads as a warm silhouette through it, one hand pressed flat against the fabric, then leans her forehead to it. Steam rolls past the candles. Slow push in, no cut.

Run this prompt on Seedance 2.5 →

Twelve more, by scene, are in our Seedance 2.5 prompt guide.

How far it goes

Seedance 2.5 went all the way up the scale we ran. On 14 and 15 September Strong and one of the two Maximum probes came back as written on all three tasks. On 24 September we added three runs per level on each task: Mild, Moderate and Strong passed every time, and Very strong passed 8 times out of 9. The one miss was a text-to-video clip that kept the scene implied rather than delivering it as asked. The model page label is Uncensored: we add no content filter on our side. Every Seedance version is compared in our Seedance uncensored test.

Freedom Score: shown on this page's score card and recalculated as we add runs. As of 24 September, one of the two Maximum probes had one run per task and the other hadn't been run, and the one Very strong clip that stayed implied is on record.

The contrast with its predecessor is sharp. On the same day, Seedance 2.0 softened 8 of its 9 Very strong runs, and its Fast and Mini versions did little better; 2.5 delivered 8 of 9. On image-to-video the still wins: its content comes through unchanged. Refusals come from Seedance's own wording check before rendering, so they cost nothing.

Not tested yet: the second Maximum probe; the word "kiss" on today's version; real photographs as stills or references; other character pairings at the upper levels.

Adults only, in every prompt, still and output. Never depict a real, identifiable person without their consent.

What the community says

The praise is about coherence. In the launch thread one commenter was "shocked by the quality and the coherence", another said you "pretty much get stellar output which is extremely consistent", and long single takes are the reason people pick it over models that drift. Even in the MiniMax H3 launch thread, one commenter rated 2.5 "significantly better" for many shots.

The complaints are specific. Faces still give it away: "stiff and unnatural, eyes aren't right". On ByteDance's own app, 2.5 launched at about twice the price of 2.0. And Seedance 2.0 launched with few filters before rights holders pushed the official apps to tighten, which is why people look for an uncensored way in.

Alternatives

  • Uncensored clips from a prompt for a fraction of the price? MiniMax H3: every clip we sent came back uncut, every level from Mild to Maximum; 5 seconds at 768p is $0.185. It stops at 15 seconds. Our H3 review.
  • Further on our record, a 4K file from a still, or the harder reading of your scene? Seedance 2.5 Spicy: it delivered every Very strong and Maximum clip we asked for on 25 September, where base 2.5 has no runs yet on the second Maximum probe; image-to-video only, 56% to 75% more per second. Our 2.5 Spicy review.
  • 30 seconds for about a third of the price? Wan 3.0 runs 30 seconds with sound from text, a still or references: $2.70 at 720p against $8.32 here. It escalates past what you wrote. Our Wan 3.0 review.
  • Native 4K? Seedance 2.0: $3.89 for 5 seconds at 4K, 15 seconds per take. Keep it for safe-for-work work: on 24 September its text-to-video turned away 7 of 8 Mild prompts as possible copyright issues and softened most Very strong ones.
  • MiniMax's fast build for safe-for-work shots? MiniMax H3 Max, $0.40 for 5 seconds at 768p. Keep upper-level text prompts off it: it answered our Very strong prompt with a toned-down result.

How far Seedance 2.5 goes

Freedom points by level and task
LevelImage to videoReference to videoText to video
MildThe lightest prompts in the test
3/3 delivered20.0 / 20 pts
3/3 delivered20.0 / 20 pts
3/3 delivered20.0 / 20 pts
ModerateWhere filtered models start to refuse
3/3 delivered20.0 / 20 pts
3/3 delivered20.0 / 20 pts
3/3 delivered20.0 / 20 pts
StrongMid-scale; must be tested before a model is ranked
7/7 delivered20.0 / 20 pts
7/7 delivered20.0 / 20 pts
2/2 delivered16.0 / 20 pts
Very strongWhere most models start to hold back
3/3 delivered20.0 / 20 pts
3/3 delivered20.0 / 20 pts
2/3 delivered16.7 / 20 pts
MaximumThe top of the scale, scored in two halves: class A and class B
Class A: 1/1 deliveredClass B: Not tested yet6.0 / 20 pts
Class A: 1/1 deliveredClass B: Not tested yet6.0 / 20 pts
Class A: 1/1 deliveredClass B: Not tested yet6.0 / 20 pts
Task score86.0 / 100.0Tested Oct 5, 202686.0 / 100.0Tested Oct 5, 202678.7 / 96.7Lowest: pulls the average downTested Sep 24, 2026

Tested Sep 14, 2026 to Oct 5, 2026

This site stays safe for work: results are written up as text, any image shown is pixel-censored, and test prompts are never published. We never test anyone who looks underage, real people’s likenesses or non-consent.

See every test record →See the uncensored leaderboard →

Run your own prompt and see how far it goes

How Seedance 2.5 earns its Spicy Index

Public leaderboards 45% of quality
83.5
Blind output test 35% of quality
65.8
Arena quality 20% of quality
Not yet tested

Counts once the model has 100+ blind votes and a 95% interval under 200 Elo. How it works

Capability 10.5% weight
89.0
Freedom Score 30% weight
83.6
Open the full test report →

Capability, item by item

Native max resolution1080p
25.0 / 25
Longest single clip30 s
14.0 / 20
Native audioYes
15.0 / 15
Reference inputs30 reference images · 20 video or audio
15.0 / 15
Frame controlFirst and last frame
10.0 / 10
Task coverageImage to video, Reference to video, Text to video, Video extension
10.0 / 10
Customization (LoRA)No LoRA support
0.0 / 5

Arena rating: 1092 (95% interval 989–1182); votes are still coming in.

What Seedance 2.5 can do on SpicyAPI

Capabilities by endpoint
FeatureImage to videoReference to videoText to videoUpscalingVideo editingVideo extension
Resolutions480p · 720p · 1080p · 4k480p · 720p · 1080p480p · 720p · 1080p · 4k1080p480p · 720p · 1080p480p · 720p · 1080p
Clip length4–30 s4–30 s4–30 s4–30 s—4–30 s
Aspect ratiosadaptive21:9 · 16:9 · 4:3 · 1:1 · 3:4 · 9:16 · adaptive21:9 · 16:9 · 4:3 · 1:1 · 3:4 · 9:16 · adaptive———
Native audioYesYesYesYesYesYes
Output formatsmp4 · movmp4 · movmp4 · movmp4 · movmp4 · mov—
Seed controlYesYesYesNoNoYes
Negative promptNoNoNoNoNoNo
See every parameter and code sample →

How long it really takes

Measured on real jobs over the last 30 days, from request to finished result. Times include our own test runs.

Median time
4.5 min
90th percentile
8.0 min
Wait per second of video
19.5 s per output second
Success rate
Collecting data

A time appears once at least 10 completed jobs are in. Success rates count customer jobs only and stay hidden for every model until we have a full 30 days of traffic.

What people say about Seedance 2.5

3 positive · 0 mixed · 2 negative

  • I'm shocked by the quality and the coherence they're able to maintain

    hacker-news· Aug 1, 2026· PositiveSource

  • for seedance you pretty much get stellar output which is extremely consistent

    hacker-news· Aug 2, 2026· PositiveSource

  • significantly better than this for a lot of cases

    hacker-news· Aug 3, 2026· PositiveSource

  • stiff and unnatural, eyes aren't right, faces move weird

    hacker-news· Aug 2, 2026· NegativeSource

  • seedance 2.5 is ~2x as expensive

    hacker-news· Aug 2, 2026· NegativeSource

Quotes are verbatim from public posts, linked to the original. They are not ratings and don’t feed any score.

Models to consider instead of Seedance 2.5

What do you think?

Was this review useful?

Log in to vote.

Comments

Comments are checked automatically and by our team. We remove spam, offensive content, personal attacks and anything that names a real person without consent. Report a comment and we review it within one business day.

Tried this model? Share what worked. You need a free account to comment, so we can keep spam out.

Seedance 2.5 FAQ

Try Seedance 2.5 yourself

Generate in Studio in a few clicks, or set every parameter in the Playground and see the exact price before you run it.

$1.39 per 5-second clip

  • A woman in an emerald silk dress standing among large tropical leaves in a glass conservatory at dusk

    Wan 3.0

    Pick Wan 3.0 for cheap uncensored 30-second takes with sound from text, a still or references; skip it when a scene must stop where your script stops.

    Spicy Index
    87.7
    Freedom Score
    96.0

    Tested Sep 24, 2026

    Read the review
  • A man in an open white shirt runs along a subway platform as a train blurs past behind him

    Seedance 2.0 Fast

    Pick Seedance 2.0 Fast for the 2.0 line's uncensored rendering from Mild to Strong without paying for 4K; at Very strong, look elsewhere in our tests.

    Spicy Index
    73.5
    Freedom Score
    72.2

    Tested Sep 24, 2026

    Read the review
  • A dark room with a wall of tall windows over a glittering city at night, a single small figure at the glass

    Wan 3.0 Prime

    Pick Wan 3.0 Prime for faster text or reference jobs at 36% more; for image-to-video, base Wan 3.0 costs less and clears the same intimate-scene level.

    Spicy Index
    75.8
    Freedom Score
    80.7

    Tested Sep 24, 2026

    Read the review