Alibaba
Wan 3.0
Wan 3.0 from Alibaba scores 4.75 out of 5 on our full 40-prompt battery, ninth of 15, with perfect marks on multi-shot sequences, text spelled correctly in all four tests and clean single takes, held back by a camera orbit that barely moves and an ice cube that never melts.
- Best category
- Multi-shot 5.00
- Weakest category
- Hands 4.37
- Clips scored
- 40
At a glance
Strengths
- Multi-shot at 5.00: the cafe in three shots, the man in the yellow jacket, the watch and the bench at midday and dusk all keep the same subject and place across the cuts.
- On-screen text is spelled correctly in all four tests, including a flickering "OPEN 24H" neon sign.
- Single takes stay single: only 3 of the 32 one-shot text-to-video prompts cut to another angle, one of them the two-person dialogue, and every text-to-video clip comes out at 1920x1080 with native audio.
Watch-outs
- The 360-degree orbit becomes a sideways slide of about 20 to 30 degrees, and the ice cube changes position instead of melting.
- Hands are its weakest category (4.37): the fingers open together instead of one by one, and the pepper reappears whole after a cut.
- The direct-to-camera line starts with a false start, "We test we the slop", before the full sentence.
- At $0.20 per second for 1080p on Alibaba Cloud, an 8-second clip lists at $1.60.
Sequences of shots that must keep the same character and place, signage and other on-screen text, and clips that need to stay in one take. Check camera moves and physical change, like melting, before relying on it.
Capabilities
- Native audio
- Yes
- Multi-shot sequences
- Yes
- Image to video
- Yes
- Public API
- Yes
- Max resolution
- 1080p
- Max duration
- 30s
Benchmark scores
How we test →| Dimension | Score | |
|---|---|---|
| Overall | 4.75 / 5 | |
| Prompt fidelity | 4.61 | |
| Visual quality | 4.89 | |
| Physics & motion | 4.50 | |
| Subject consistency | 4.97 | |
| Audio & sync | 4.75 |
- Scored runs
- 40
| Prompt category | Score | Runs | |
|---|---|---|---|
| Multi-shot | 5.00 | 4 | |
| On-screen text | 4.92 | 4 | |
| Faces | 4.92 | 4 | |
| Image to video | 4.88 | 4 | |
| Style | 4.79 | 4 | |
| Animals | 4.79 | 4 | |
| Lip-sync | 4.75 | 4 | |
| Camera | 4.67 | 4 | |
| Liquids | 4.42 | 4 | |
| Hands | 4.37 | 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.
SlopTV. (2026). Wan 3.0 benchmark results. Retrieved October 21, 2026, from https://sloptv.co/model/wan-3-0<a href="https://sloptv.co/model/wan-3-0">Wan 3.0 benchmark results</a> (SlopTV)The clips
Every score above links to the actual generation it came from. Nothing here is a mockup.
Hands avg 4.37/5
Chopping a pepper
Clean slices and two hands throughout, but a cut at about 2 seconds puts the pepper back together whole and swaps which hand holds the knife. Verified frame by frame.
Counting fingers
Five digits throughout, checked frame by frame. The thumb opens first and the other four fingers then straighten together instead of one by one, and the fingertips leave the top of the frame once the palm is open.
Fretting a guitar chord
Fingertips press the strings and shift between chords, with five fingers on the fretting hand throughout.
Tying shoelaces
Both hands keep five fingers while the laces cross, and the bow is finished by the end of one continuous take.
Faces avg 4.92/5
Face in a crowd
The man stays still and sharp, but the crowd behind him is only slightly softened rather than out of focus. Checked frame by frame.
Elderly portrait, slow blink
Deep wrinkles hold under hard side light and there is one slow, complete blink.
Glasses reflection
Round glasses carry a window reflection that moves with the slight head motion, without warping the frames or the eyes.
Turn and smile
Turns from profile to camera and breaks into a smile showing teeth, with the same face and hair through the turn.
Liquids avg 4.42/5
Pouring coffee
Continuous slow-motion pour into clear glass, with the level rising and backlit steam, in one take.
Honey drizzle
Thick, slow ribbons from a wooden dipper that pool on the stack and start to run over the edge.
Ice melting in whisky
Condensation builds on the glass, but the cube does not visibly shrink: at about 1.5 seconds it snaps from a tilted to an upright position and changes shape. Verified frame by frame.
Wave on rocks
The wave breaks on rigid dark rocks and the spray falls back under gravity, under an overcast sky.
On-screen text avg 4.92/5
Book chapter heading
"CHAPTER ONE" is spelled correctly and stays stable during the push-in. The body text on the page is pseudo-text.
Neon sign OPEN 24H
"OPEN 24H" reads correctly throughout, and the sign dims and relights several times on a wet street at night.
Bakery sign
"FRESH BREAD DAILY" reads correctly on the painted sign in morning light. A chalkboard inside carries illegible pseudo-text.
Printed t-shirt
"SLOP" stays legible on the white t-shirt, following the fabric folds, as the person walks toward the camera.
Multi-shot avg 5/5
Cafe, three shots
Three shots in the asked order, wide, medium, then hands around the cup, with cuts at about 2.5 and 4.5 seconds and the same woman, red coat and cafe. Checked frame by frame.
Same bench, day and dusk
Midday sun with hard shadows, then dusk after a cut at about 3.5 seconds, on the same bench with identical framing. Checked frame by frame.
Watch, wide then macro
Wide on the stand, then a macro of the same blue-dial watch with the same bracelet, hands and date window. Checked frame by frame.
Alley, two angles
From behind down the alley, then front-on after a cut at about 3.5 seconds, same man and same yellow jacket. Checked frame by frame.
Lip-sync avg 4.75/5
Bearded man, long line
Audio says the exact line, transcribed word for word, and the bearded jaw moves with it without warping.
Direct address to camera
Says "We test we the slop" before delivering the full line, transcribed word by word; the lips follow both attempts.
Sung line
Sung "hold on to me", confirmed in the audio, with long held vowels and visible breath.
Two-person exchange
The man asks and the woman answers in a close-up, both lines transcribed word for word, and the listener's mouth stays still.
Camera avg 4.67/5
Crane to skyline
Smooth rise that ends on a golden-hour skyline, but it starts a few storeys above the street rather than at street level. Checked frame by frame.
Dolly in to a door
Steady push-in down a dim hallway toward a door that stays closed, centred and in proportion.
Handheld follow
Natural handheld shake following a child from behind through tall grass, with a coherent running motion.
360 orbit
The camera slides about 20 to 30 degrees across the front of the seated woman instead of orbiting her; she is never seen from the side or the back. Verified frame by frame.
Animals avg 4.79/5
Pigeon takeoff
The pigeon leaves the ledge cleanly with wings fully extended at the top of the stroke.
Cat jumping to a counter
Take-off from the floor and a natural landing, but the leap is split by a cut to a new angle at about 2.3 seconds rather than shown as one continuous motion. Verified frame by frame.
Dog shaking off water
Wet fur and droplets flying outward, though the shake is concentrated in the head and ears more than the whole body. Checked frame by frame.
Galloping horse
Side profile with four distinct legs and a gallop that includes the moment with all hooves off the ground. Checked frame by frame.
Style avg 4.79/5
Anime walk cycle
Clean anime style with face and outfit unchanged, but the camera tracks her walk, so she stays in the centre instead of crossing the frame. Checked frame by frame.
Claymation wave
Reads as clay stop-motion, with fingerprints and tool marks on the figure as it raises a hand and waves.
1970s film look
Warm 1970s grade and period car on a coastal road with halation near the horizon; the grain is light.
Rotating mug
Plain white mug turns continuously and the handle travels all the way around without jumping or duplicating.
Image to video avg 4.88/5
Animate a landscape
Clouds drift and grass sways while the terrain stays fixed, with no camera move. Compared with the reference image.
Animate a vintage photo
Stays sepia with the original grain, but adds a marked push-in toward the family instead of only subtle motion. Compared first and last frames.
Animate a portrait
Same face from first to last frame, with three natural blinks and slight hair motion. Compared with the reference image.
Product photo to spin
Turns about 180 degrees, handle from the right through the front to the left, with rigid geometry. Output keeps the square format of the reference image. Verified frame by frame.