Kling is Kuaishou's family of AI models, and three of them work here: Kling 3.0 Pro for video, Kling O3 for pictures and Kling AI Avatar v2 for talking cards. Kling 3.0 Pro takes one picture of your people and turns it into a 1080p clip of up to 15 seconds, with sound or without - the longest take in our video picker.
Updated 29.09.2026
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
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
Happy birthday! I wish I could hug you today - this is the next best thing.
MiniMax H3 is what Auto picks for video, and every one of the 20 video scenes in the catalogue was animated with it. It is the best value here and always comes with sound, so for a cheerful five-second clip it is usually enough. Kling 3.0 Pro costs more per second and earns it in two places: longer takes, and movement that has to stay smooth from the first frame to the last.
Think of a moment with an arc. The kids run down the path, Grandpa turns, they collide in a hug, he nearly spills his tea. That needs eight or ten seconds, sometimes fifteen, and motion that holds together while people cross the frame. The picnic clip above is the short version: paper hats, party horns, a burst of confetti and a hug for the birthday girl, all in five seconds with sound.
Two honest exceptions. Veo 3.1 from Google has the cinema look, and it is the model to try when voices or singing carry the clip - it costs more per second and stops at 8 seconds. Wan 3.0 Prime from Alibaba is tagged Steady loops: 5 or 10 seconds of a gentle scene that can play on repeat.
Kling 3.0 Pro makes clips of 5, 8, 10 or 15 seconds, always in 1080p. The rest of our video models stop at 10 seconds or sooner. Fifteen is long enough for Dad to spin Mum round the kitchen, stop, and laugh about it.
Its strength here is movement: glasses rising together, a hug that lands, confetti drifting down. In the garden table clip above, the whole family raises a glass, Grandma wipes away a happy tear and the leaves stir in the breeze - five seconds, one continuous take.
Kling 3.0 can make sound in the same generation as the video: clinking glasses, party horns, laughter, birdsong. Switch sound off for a quiet moment and the clip costs less per second. Either way, the price is on the button before you press it.
Kuaishou's models do more than video here. Kling O3 is one of our cheapest picture models, handy for trying an idea before you spend more. Kling AI Avatar v2 makes a portrait speak your greeting, face and head moving with the words, for up to 15 seconds of speech.
My parents slow-dance in their small kitchen after dinner, Dad spins Mum once and she laughs into his shoulder, a warm lamp over the table, the camera circles them slowly, soft radio music and dishes clinking in the sink
Three grandchildren in paper crowns run down the garden path and hug Grandpa, he nearly drops his cup of tea and laughs, late afternoon sun, the camera follows at child height, sounds of running feet, giggles and a blackbird
Make a talking card from Grandma Rosa's photo. She says: Happy birthday, Leo! Eight already. Save me the piece of cake with the most icing.
Pick Kling 3.0 Pro when the moment needs motion and time, Kling O3 when you want to try picture ideas cheaply, and Kling AI Avatar v2 when one face should say the words. For a cheaper everyday clip there is MiniMax H3; for the cinema look, Veo 3.1.
| Model | Makes | Strong at | Length | Quality | Sound | Price |
|---|---|---|---|---|---|---|
| Kling 3.0 ProKuaishou | Video | Smooth motion | 5, 8, 10, 15 s | 1080p | With or without sound | from 700 credits · 210 cr a second |
| Kling O3Kuaishou | Photo cards | Budget | up to 8 photos of your people | 1K, 2K | from 40 credits | |
| Kling AI Avatar v2Kuaishou | Talking cards | Face and head move with the words | up to 15 s of speech | 71 credits a second of speech + the voice |
Prices are in credits and are always shown on the button before you press it. Pricing
The button on this page opens the composer in Standard mode, on Video, with Kling 3.0 Pro already picked in the Model picker. If you switch formats or start from the chat, check the picker: left on Auto, it chooses MiniMax H3.
Kling animates a single start picture: the first photo you add, or a card you made here earlier. Add JPG, PNG or WebP files up to 10 MB from your device, your library or People. With no picture at all, it makes the clip from your words alone.
Say who moves, what they do and what we should hear. Pick 5, 8, 10 or 15 seconds and sound on or off, and watch the price update on the button. Send the finished clip by email, SMS or popular messengers, now or scheduled for the day.
Kling 3.0 Pro starts from one picture and moves it forward in time. What is in that first frame is what you get: if Aunt Dina is missing from the photo, she will not stroll in later looking like herself. So the best Kling videos here are made in two steps - first a photo card with everyone in it, then the animation.
For that card, GPT Image 2.5 Flare keeps faces most like the real people and is what Auto uses for pictures. Nano Banana Pro holds a likeness for up to 5 people, which suits a long family table. One picture can hold up to 6 people, but faces stay most reliable up to 4.
Kling O3, Kuaishou's picture model, is one of the cheapest here. Use it to test a setting - the beach or the balcony, morning or dusk - then switch to Flare for the version that has to look like Mum. Once the card is right, bring it to life with Kling 3.0 Pro.
The talking card is the other Kling you will meet here. You give it one portrait and your greeting, pick a voice, and Kling AI Avatar v2 makes the face and head follow the words, for up to 15 seconds of speech. In the example above, a grandmother looks into the camera and says: "Happy birthday! I wish I could hug you today - this is the next best thing." That line takes 7 seconds in the Ava voice, so 15 seconds fits about two lines like it.
The voices come from ElevenLabs Eleven v3: 8 catalogue voices, Hebrew included - no other voice model here speaks it. Kling AI Avatar v2 is the Better option in the talking card settings. Best is OmniHuman 1.5 from ByteDance, which moves the whole person, not only the face, and lets you point at who speaks when several people share the photo. It costs more per second of speech, so for a single close-up portrait Kling is usually the sensible pick.
Kling 3.0 Pro is Kuaishou's video model, and of the three Kling models we run it is the only one…
OpenVeoVeo is Google's video model, here in three versions: Veo 3.1, Veo 3.1 Fast and Veo 3.1 Lite. Each turns…
OpenMiniMaxMiniMax H3 is the AI video model from MiniMax: it turns a photo card, or a few lines about the…
OpenTalking photo cardA talking photo card has someone you love say your greeting out loud - up to 15 seconds of speech,…
OpenAI video greetingAn AI video greeting starts from a photo of your people and ends as a short clip where they move…
OpenKling O3 is Kuaishou's picture model, and here it is tagged Budget: one of the cheapest ways to make a still card, at 1K or 2K, with up to 8 reference photos. Use it to try a setting first, the beach or the balcony, morning or dusk, before you spend more. For the version where Mum has to look like Mum, switch to GPT Image 2.5 Flare.
Up to 15 seconds. Kling 3.0 Pro makes clips of 5, 8, 10 or 15 seconds, always in 1080p, and 15 is the longest length in our video picker. Video is priced per second, so longer clips cost more. For a quick five-second moment, MiniMax H3 is the cheaper choice.
Yes. Kling 3.0 Pro can make sound in the same generation as the video, so the horns and the cheering in the picnic clip above came with the picture. Sound is a switch in the settings; turn it off and each second costs less. Both Kling 3.0 clips on this page were made with sound on, 5 seconds each, in 1080p.
Yes. Add photos of your people as JPG, PNG or WebP files up to 10 MB, from your device, your library or People. Kling 3.0 Pro animates one start picture, so for a group, first make a photo card with everyone in it, then turn that card into a clip. Faces stay most reliable with up to 4 people in the picture.
Kling AI Avatar v2 is Kuaishou's talking photo model. Here it powers the Better talking card: a portrait speaks your greeting in the voice you pick, with the face and head moving to the words, for up to 15 seconds of speech. If you want the whole person to move, or several people share the photo, choose OmniHuman 1.5.
Kling 3.0 Pro is the better pick for longer takes, up to 15 seconds, and smooth movement in 1080p. Veo 3.1 is the one for the cinema look and for clips where voices or singing matter, though it costs more per second and stops at 8 seconds. If you are unsure, Auto uses MiniMax H3, the best value here.
Pick your people, describe the moment and send the clip on the day by email, SMS or popular messengers.
Make a Kling video