Fifteen seconds for the whole toast: Kling 3.0 from one photo
Kling 3.0 Pro is Kuaishou's video model, and of the three Kling models we run it is the only one that sets a whole scene in motion. Start it from one photo of your people and it plays the moment on for 5, 8, 10 or 15 seconds at 1080p: people get up, glasses meet, somebody gets hugged, with the sound of it or in silence.
- Made by Kuaishou
- from 700 credits
- 5-15 s
- With or without sound
Updated 29.09.2026
Two 5-second Kling 3.0 clips, both with sound
Prompt
The family around the long garden table raise their glasses together and laugh, grandmother at the head of the table smiles and wipes a happy tear, leaves move in a light breeze, warm golden light, natural gentle camera push-in, sounds of clinking glasses, laughter and birdsong
Prompt
Friends at the picnic in paper party hats blow party horns and toss confetti into the air, one of them laughs and hugs the birthday girl, confetti floats down slowly, handheld camera feel, sunny park, sounds of party horns, cheering and laughter
What Kling 3.0 does that its Kling siblings here don't
People leave their seats
Kling AI Avatar v2 moves a face and a head; Kling O3 makes stills. Kling 3.0 moves people through the frame. In the picnic clip above, by the fourth second the friend in the denim jacket is up off the blanket and hugging the birthday girl, confetti still in the air.
The longest take in our picker
Choose 5, 8, 10 or 15 seconds. No other video model here goes past 10, and Veo 3.1 stops at 8. Think in beats: five seconds holds one gesture, ten holds a gesture and a reaction, fifteen holds a small story with a start and an end.
Faces guarded on every request
Your photo is the first frame, so the clip starts with your actual people. With every Kling 3.0 request we also send a standing instruction to avoid morphing faces, extra people and stray text. In the garden-table clip, all four people still look like themselves when their glasses meet.
Sound you can switch off
MiniMax H3, our Auto for video, always comes with sound. Kling 3.0 lets you choose: clinks, laughter and birdsong made in the same generation as the picture, or a silent clip that costs less per second. Silent is the right call when the clip travels with a song you made here.
Making a Kling 3.0 clip: picture, length, sound
- 1
Pick the first frame
Use the button on this page, which opens Video with Kling 3.0 Pro already set, or switch the composer to Standard and choose it under Model. Attach one picture: a photo of your people (JPG, PNG or WebP up to 10 MB) or a card you made here.
- 2
Choose a length, then write the beats
Decide on 5, 8, 10 or 15 seconds first, because the length sets how much can happen. Then write the moment in order: who moves first, what happens next, how it ends, and last of all, what we should hear.
- 3
Set sound, check the price, send
Leave sound on or switch it off; the button shows the new price after every change, and a failed clip returns its credits. The finished clip goes out by email, SMS or a popular messenger, straight away or on the morning of the day.
Three Kling models here, three different jobs
Kuaishou makes all three Kling models we run, and they barely overlap. Kling O3 draws still pictures and is one of the cheapest picture models in our picker, handy for trying out a setting. Kling AI Avatar v2 is the engine of our talking card: one portrait, your text, a voice you choose, up to 15 seconds of speech. Kling 3.0 Pro is the only one of the three that animates a whole scene: several people, the camera, the room and the sound of it.
The line between them matters most when words are involved. Kling 3.0 makes the sounds of a moment, the clinking and the laughter, but it does not read out the greeting you typed. When Grandma should say "Happy birthday, Leo" herself, you need a talking card, and the Better option there is Kling AI Avatar v2.
They also work well one after another. Make the picture of your people first (for faces, GPT Image 2.5 Flare, our Auto for pictures, keeps the likeness best), bring it to life with Kling 3.0, and if the clip is silent, send a song made here with Lyria 3.5 alongside it.
Does the face hold? What our two clips show
The garden-table clip starts from one of our scene pictures: four people at a table under fruit trees, a birthday cake in the middle. Over 5 seconds the camera pushes in and all four glasses meet over the cake. Each face stays the same person from the first frame to the last, including the younger woman who turns away from us to reach in.
The picnic clip is the harder test. Four friends in paper hats blow party horns, then one jumps up and crosses the blanket to hug the birthday girl. For a moment in mid-jump she blurs, as anyone would on a phone camera, and then the two of them settle into the hug as the same people who started the clip.
What slips is whatever the first frame never showed. Someone who walks in from outside the picture is invented, because the model has never seen them, and a tiny face at the far end of a table gives it little to hold on to. So start from a picture with everyone already in it, faces clear and in the light. If your people are in separate photos, make the card first: Flare for up to four people, Nano Banana Pro for five.
Kling 3.0 prompts, one for each length
Adam leans over his cake and blows out all six candles in one go, his big sister claps and hugs him from behind, smoke curls up from the candles, warm kitchen light, sounds of blowing, clapping and a small cheer
The two kids on the sofa jump up and throw themselves at Dad, he pretends to be squashed and falls back into the cushions, then all three burst out laughing, the dog barks once, warm lamp light, camera at sofa height
Grandpa at the head of the long table taps his glass with a fork, everyone turns to him, he stands and lifts the glass, then the whole table stands and glasses meet all the way down, Grandma pulls him down for a kiss on the cheek, evening light, the camera slowly pulls back to show the whole table, sounds of the fork on glass, chairs scraping and cheering
Kling 3.0 Pro in numbers, priced live
These are the settings Kling 3.0 Pro runs on here today and what a second costs, with sound and without. For a five-second clip that always has sound, MiniMax H3 costs less; for a spoken line or a chorus, look at Veo 3.1.
| Model | Strong at | Length | Quality | Sound | Price |
|---|---|---|---|---|---|
| Kling 3.0 ProKuaishou | Smooth motion | 5, 8, 10, 15 s | 1080p | With or without sound | from 700 credits · 210 cr a second |
Prices are in credits and are always shown on the button before you press it. Pricing
Keep exploring
Kling is Kuaishou's family of AI models, and three of them work here: Kling 3.0 Pro for video, Kling O3…
OpenVeo 3.1Veo 3.1 is the full model of Google's Veo line and the top of the three Veo options in our…
OpenMiniMaxMiniMax H3 is the AI video model from MiniMax: it turns a photo card, or a few lines about the…
OpenAI video greetingAn AI video greeting starts from a photo of your people and ends as a short clip where they move…
OpenQuestions and answers
Mostly, yes: your photo is the first frame, and every Kling 3.0 request we send asks it to avoid morphing faces and extra people. In our garden-table clip all four faces hold for the full 5 seconds. Likeness slips with big turns, very small faces and anyone who walks in from outside the picture, so start from a clear photo with everyone in it.
Kling 3.0 Pro is the version of Kling 3.0 we run, and it appears under that name in the Model picker. Every clip it makes here is 1080p. There is no second Kling 3.0 option to weigh up: what you choose is the start picture, the length and whether the clip has sound.
No, that is a different Kling model. Kling 3.0 makes the sounds of a moment, like clinking, laughter or party horns, not your typed words in a voice you picked. For that, make a talking card: Kling AI Avatar v2 moves her face and head as she speaks, for up to 15 seconds, and OmniHuman 1.5 moves her whole figure.
About three beats. Each of our 5-second clips holds one: glasses meeting over a cake, or a jump up and a hug. Fifteen seconds lets a moment start, turn and end, like a toast that grows from one tap on a glass to the whole table on its feet. Write the beats in order and keep everyone who moves in the first frame.
Turn it on when the room tells the story: clinks, cheering, horns, birds in the garden. Kling 3.0 makes that sound in the same generation as the picture. Turn it off when the clip will go out with a song or be watched on a muted phone, because a silent second costs less.
For a quick clip of 5 to 10 seconds, MiniMax H3: it is what Auto picks, it costs less per second, it always has sound, and every video scene in our catalogue was made with it. Pick Kling 3.0 Pro when you need more than 10 seconds, a silent version, or a lot of people moving at once in 1080p.
Give the moment time to turn
Pick one photo of your people, write the beats in order, and let Kling 3.0 play it out for up to 15 seconds.
Start a Kling 3.0 clip
