ByteDance
Seedance 2.0
ByteDance Seedance 2.0 measured against our full 40-prompt battery scores 4.84 out of 5, level with Kling 3.0 Turbo and FLUX 3 for first of 16 models. Its weakest category is hands, where pepper slices duplicate on the chopping prompt.
- Best category
- Faces 5.00
- Weakest category
- Hands 4.09
- Clips scored
- 39
- Generation time
- 267s
What our own testing found
We ran the same 40-prompt battery we run against every model in this category, at 1080p, 8 seconds, 16:9. Seedance 2.0 averaged 4.84 across 39 usable clips, level with Kling 3.0 Turbo and FLUX 3 for first of the 14 models we have measured, ahead of Sora 2 at 4.76, Grok at 4.72 and Kling 3.0 at 4.47.
It produced native audio on every clip we generated, not just the four lip-sync prompts, which nobody else in this category does by default.
One real failure, not five stars across the board
The vegetable-chopping prompt duplicated pepper slices mid-cut. The hand-counting prompt kept five digits in every frame, but the four fingers opened together in one wave instead of one at a time. A third weak spot was stylistic rather than a hard failure: asked for 1970s film grain, it defaulted to a clean modern digital look with no period grain or halation.
One prompt (a man standing in a crowded train station) was blocked twice by the platform's own moderation rules before we could generate it at all, so it carries no score.
What we could not confirm from outside
Third-party rankings citing Seedance 2.0 at the top of the field (commonly a 4.70/5 figure) were based on a six-prompt sample from an April 2026 snapshot that has not been refreshed since. Our own 40-prompt measurement is the most complete public data point we are aware of for this model.
At a glance
Strengths
- Joint first of 14 models overall at 4.84/5, with perfect scores on subject consistency and audio sync.
- Native audio generation built in.
Watch-outs
- Physics scored 4.59, its weakest dimension, and below what Seedance 2.5 scores on the same test.
Prompts that need reliable audio sync and consistency across the clip more than they need top-tier physics.
Capabilities
- Native audio
- Yes
- Multi-shot sequences
- No
- Image to video
- Yes
- Public API
- Yes
Benchmark scores
How we test →| Dimension | Score | |
|---|---|---|
| Overall | 4.84 / 5 | |
| Prompt fidelity | 4.79 | |
| Visual quality | 4.90 | |
| Physics & motion | 4.59 | |
| Subject consistency | 5.00 | |
| Audio & sync | 5.00 |
- Average generation time
- 267s
- Scored runs
- 39
| Prompt category | Score | Runs | |
|---|---|---|---|
| Faces | 5.00 | 3 | |
| On-screen text | 5.00 | 4 | |
| Multi-shot | 5.00 | 4 | |
| Lip-sync | 5.00 | 4 | |
| Camera | 5.00 | 4 | |
| Animals | 5.00 | 4 | |
| Image to video | 5.00 | 4 | |
| Style | 4.75 | 4 | |
| Animation | 4.72 | 4 | |
| Liquids | 4.63 | 4 | |
| Hands | 4.09 | 4 |
Scores out of 5, judged blind. Bars start at 3.0 so the gaps are visible.
Cite this
Free to cite and reuse with a link back. Data is updated as new runs and prices come in, so include the date.
Genativity. (2026). Seedance 2.0 benchmark results. Retrieved October 23, 2026, from https://genativity.com/model/seedance-2-0<a href="https://genativity.com/model/seedance-2-0">Seedance 2.0 benchmark results</a> (Genativity)The clips
Every score above links to the actual generation it came from. Nothing here is a mockup.
Hands avg 4.09/5
Chopping a pepper
FAILURE MODE: pepper slices duplicate and appear simultaneously in front of the blade instead of separating one cut at a time.
Counting fingers
Five digits throughout. The thumb opens first, then the four fingers unfold together in one wave instead of one at a time.
Fretting a guitar chord
Natural contact with the strings and fretboard, but the guitar neck has extra frets.
Tying shoelaces
Ten normal digits across both hands, no clipping or melting into laces or leather.
Faces avg 5/5
Elderly portrait, slow blink
Genuine full blink with wrinkles and skin texture preserved, not smoothed away.
Glasses reflection
Window reflection tracks correctly and plausibly with head movement.
Turn and smile
Facial identity holds through the full turn into the smile, no reset mid-motion.
Liquids avg 4.63/5
Pouring coffee
Continuous pour with plausible refraction and steam, no looping artifacts.
Honey drizzle
Visibly high viscosity, thick ribboned flow clearly distinct from water.
Ice melting in whisky
FAILURE MODE: essentially a static render. No visible melting or shrinking of the ice, no new condensation forming over the 8 seconds.
Wave on rocks
Chaotic, non-repeating spray and foam on impact.
On-screen text avg 5/5
Book chapter heading
Reads "CHAPTER ONE" exactly, stays locked to the page through the push-in.
Neon sign OPEN 24H
Reads "OPEN 24H" exactly, stable through a brief flicker.
Bakery sign
Reads "FRESH BREAD DAILY" exactly across all three lines.
Printed t-shirt
Reads "SLOP" exactly, stays legible as the fabric deforms with the walk.
Multi-shot avg 5/5
Cafe, three shots
Same woman, coat, mug and table hold identical across all three shots.
Same bench, day and dusk
Framing and bench geometry stay locked while lighting genuinely shifts from midday to dusk.
Watch, wide then macro
Watch design details match exactly between the wide and macro shots.
Alley, two angles
Same jacket, build and alley geometry hold across the cut to the front angle.
Lip-sync avg 5/5
Bearded man, long line
Audio present, mouth shapes match despite the beard partially hiding the jaw.
Direct address to camera
Audio present, mouth shapes match the stated line.
Sung line
Audible singing with sustained, elongated vowel shapes distinct from normal speech.
Two-person exchange
Both lines are audible and matched to the right speaker; the listening partner keeps a still mouth in each turn.
Camera avg 5/5
Crane to skyline
Real reveal: skyline is hidden at frame one and becomes visible only as the crane rises.
Dolly in to a door
Perspective changes correctly on approach, no warping or drift on the door.
Handheld follow
Controlled natural shake rather than a motorized glide.
360 orbit
Completes a genuine continuous 360-degree orbit with the room geometry holding throughout.
Animals avg 5/5
Pigeon takeoff
Wings articulate correctly and feet cleanly leave the ledge.
Cat jumping to a counter
Full ballistic arc with a real crouch, push-off and momentum-absorbing landing.
Dog shaking off water
Plausible high-frequency fur and droplet dynamics.
Galloping horse
Plausible gallop sequence with a real suspension phase.
Style avg 4.75/5
Anime walk cycle
Face and outfit stay consistent across the walk.
Claymation wave
Visible clay texture and tool marks, not smooth CGI.
1970s film look
FAILURE MODE: defaults to a clean modern digital look. No period-accurate grain or lens halation, despite an otherwise stable, well-rendered shot.
Rotating mug
Handle travels smoothly around the full rotation, no teleporting or duplication.
Image to video avg 5/5
Animate a landscape
Only clouds and grass move; the terrain stays static.
Animate a vintage photo
Motion added without colourising or sharpening; vintage tonality preserved.
Animate a portrait
Source facial identity fully preserved through the animation.
Product photo to spin
Plausible, consistent geometry invented for the unseen sides of the mug.
Animation avg 4.72/5
3D feature film look
The clearest acting of the set: the robot tilts its head, its eyes widen, it slowly lifts a hand and the butterfly settles on it at about 5.5 seconds. Soft forest light and blush glow on the cheeks.
Anime sword draw
Cinematic rather than static: a crouch, a glowing draw, a tracking side-view lunge with long trailing hair, then a cut at 4.5 seconds to an extreme wide shot where she is tiny for the last three seconds.
Squash and stretch
Clean anticipation, stretch while falling, squash on landing and a surprised face at the end, but the ball is a glossy 3D model in a photographic room rather than flat 2D.
2D walk cycle
An even two-legged walk in place with arm swing and a stable design. The character reads more like a cat than a fox.
Where to use it
The same model costs different amounts depending on the platform you reach it through.
| Platform | 10s costs | Max res | Watermark | Commercial use | |
|---|---|---|---|---|---|
 Higgsfield |
$3.1 | 4K | no | yes | Visit |
 Runway |
$7.68 | 1080p | no | yes | Visit |
 Magnific |
— | 4K | no | yes | Visit |
 LTX Studio |
— | — | no | yes | Visit |
 Leonardo.Ai |
— | — | — | — | Visit |
 ImagineArt |
— | 4K | — | yes | Visit |
 Krea |
— | — | — | yes | Visit |
 OpenArt |
— | 4K | no | yes | Visit |
 Dreamina |
— | 720p | — | — | Visit |
 Lumeta |
— | 4K | no | yes | Visit |