MiniMax Hailuo Video Generator

4 to 15-second clips in 2K with sound. Start from a sentence, a first and last frame, or up to 9 references.

Minimax H3Starter
Generate

On Starter · A 4-second 768P clip is 3.5 credits

Minimax H3

Warm handheld family footage in a cosy home kitchen on a Sunday afternoon. A grandmother in her seventies in a floral apron sits at the kitchen table with her teenage grandson; a pot of soup steams on the stove behind them. He tells her something and she bursts out laughing, swatting his arm gently. She turns to the camera, still laughing, and says: "He thinks he can cook better than me." Soft afternoon light, kitchen clock ticking. Authentic footage shot on a phone, natural daylight, real skin texture with pores, unretouched, slight handheld movement, realistic ambient sound, no music. Hands stay relaxed; nothing is opened, sprayed, poured or assembled. No logos, no brand names, no readable text anywhere. No on-screen text, no captions, no subtitles.

Minimax H3

Warm handheld family footage in a cosy home kitchen on a Sunday afternoon. A grandmother in her seventies in a floral apron sits at the kitchen table with her teenage grandson; a pot of soup steams on the stove behind them. He tells her something and she bursts out laughing, swatting his arm gently. She turns to the camera, still laughing, and says: "He thinks he can cook better than me." Soft afternoon light, kitchen clock ticking. Authentic footage shot on a phone, natural daylight, real skin texture with pores, unretouched, slight handheld movement, realistic ambient sound, no music. Hands stay relaxed; nothing is opened, sprayed, poured or assembled. No logos, no brand names, no readable text anywhere. No on-screen text, no captions, no subtitles.

Length
4 to 15s
Resolution
768P or 2K
Aspect ratio
21:9 to 9:16
Audio
Built in
Start from
Text, frames or references
Minimax H3
Starter · 3.5 credits for 4s

Create with Minimax H3

  1. Minimax H3Starter
    Generate
    01

    Pick Minimax H3

    Open Model Playground and choose Minimax H3: write the scene, start from a first and last frame, or add references.

  2. Minimax H3Starter
    Generate
    02

    Pick the length and resolution

    4 to 15 seconds at 768P or 2K. Sound comes in the same render; a 4-second 768P clip is 3.5 credits.

  3. 03

    Generate

    Minimax H3 renders the clip with sound in Model Playground, ready to download.

Minimax H3Starter
Generate

See it in action

Whether you are making a product ad, a short film or a post for social, Minimax H3 on Creatify gets you from a sentence to the shot.

For marketers: footage that feels filmed

A grandmother laughs with her grandson in a Sunday kitchen, then turns to camera: “He thinks he can cook better than me.” The line, the laugh and the room sound come out of one 2K render.

For filmmakers: a car spot in 21:9 you couldn’t permit

A coupe racing a lightning storm across a mirror salt lake, in 2K widescreen with its sound made in the same render, from one paragraph of text.

For fashion brands: one take through four worlds

A silk gown walks from desert dunes through a snowstorm and a rainy rooftop to a sunrise beach, changing colour in each, in one unbroken 15-second shot.

The prompts behind these clips

Every clip on this page came out of Minimax H3 in Creatify with these exact prompts; we only compressed them for the web and cut the park walk to its first five seconds. Copy one and make it yours.

  • Warm handheld family footage in a cosy home kitchen on a Sunday afternoon. A grandmother in her seventies in a floral apron sits at the kitchen table with her teenage grandson; a pot of soup steams on the stove behind them. He tells her something and she bursts out laughing, swatting his arm gently. She turns to the camera, still laughing, and says: "He thinks he can cook better than me." Soft afternoon light, kitchen clock ticking. Authentic footage shot on a phone, natural daylight, real skin texture with pores, unretouched, slight handheld movement, realistic ambient sound, no music. Hands stay relaxed; nothing is opened, sprayed, poured or assembled. No logos, no brand names, no readable text anywhere. No on-screen text, no captions, no subtitles.

    Warm handheld family footage in a cosy home kitchen on a Sunday afternoon. A grandmother in her seventies in a floral apron sits at the kitchen table with her teenage grandson; a pot of soup steams on the stove behind them. He tells her something and she bursts out laughing, swatting his arm gently. She turns to the camera, still laughing, and says: "He thinks he can cook better than me." Soft afternoon light, kitchen clock ticking. Authentic footage shot on a phone, natural daylight, real skin texture with pores, unretouched, slight handheld movement, realistic ambient sound, no music. Hands stay relaxed; nothing is opened, sprayed, poured or assembled. No logos, no brand names, no readable text anywhere. No on-screen text, no captions, no subtitles.

  • Epic car commercial, anamorphic 21:9, dusk turning from gold to deep blue. A sleek low grand-touring coupe with a plain unmarked matte graphite body, no badges, no logos, no number plates and no lettering anywhere, crosses a mirror-flat salt lake. [0-3s] One continuous drone move drops from high above the lake, where the car is a tiny dot leaving a thin trail on the mirror surface, down to wheel height beside it at speed. [3-7s] Low tracking shot along the flank: behind the car a towering thunderstorm front rolls in, lightning forks reflected in the wet salt, the car's reflection racing beneath it. [7-10s] The coupe punches through a curtain of falling rain into a band of last sunlight, water spraying off the tyres in slow motion, as the camera rises into a wide panorama of storm and sun. Sound: a deep engine growl rising to a roar, rolling thunder, tyres hissing on wet salt, a cinematic boom as it breaks into the light. Photoreal, rich contrast, lens flares. No on-screen text, no logos, no captions, no subtitles.

    Epic car commercial, anamorphic 21:9, dusk turning from gold to deep blue. A sleek low grand-touring coupe with a plain unmarked matte graphite body, no badges, no logos, no number plates and no lettering anywhere, crosses a mirror-flat salt lake. [0-3s] One continuous drone move drops from high above the lake, where the car is a tiny dot leaving a thin trail on the mirror surface, down to wheel height beside it at speed. [3-7s] Low tracking shot along the flank: behind the car a towering thunderstorm front rolls in, lightning forks reflected in the wet salt, the car's reflection racing beneath it. [7-10s] The coupe punches through a curtain of falling rain into a band of last sunlight, water spraying off the tyres in slow motion, as the camera rises into a wide panorama of storm and sun. Sound: a deep engine growl rising to a roar, rolling thunder, tyres hissing on wet salt, a cinematic boom as it breaks into the light. Photoreal, rich contrast, lens flares. No on-screen text, no logos, no captions, no subtitles.

  • One continuous 15-second single take with no cuts, luxury fashion film. A tall model with short platinum hair in a long flowing silk gown walks straight toward the camera as the camera glides backward in front of her, matching her pace. [0-4s] She walks across red desert dunes at noon, the gown burnt orange, silk trailing in the hot wind. [4-8s] Mid-stride a snowstorm sweeps over her and the world turns into a white frozen plain; the gown ripples into icy silver as snow swirls around it. [8-12s] The snow becomes rain and she is walking across a wide empty rooftop at night, wet and mirror-like, with only soft out-of-focus city lights far away and no buildings close by; the gown is now deep sapphire and reflects in the puddles. [12-15s] The rain clears to a sunrise beach, the gown turns soft gold, she stops, looks into the lens, and the silk billows out behind her. Sound: one continuous elegant electronic beat, desert wind, a snowstorm whoosh, rain on asphalt, gentle waves at the end. Photoreal, editorial lighting. The gown has no logo, no monogram, no text. No on-screen text, no captions, no subtitles.

    One continuous 15-second single take with no cuts, luxury fashion film. A tall model with short platinum hair in a long flowing silk gown walks straight toward the camera as the camera glides backward in front of her, matching her pace. [0-4s] She walks across red desert dunes at noon, the gown burnt orange, silk trailing in the hot wind. [4-8s] Mid-stride a snowstorm sweeps over her and the world turns into a white frozen plain; the gown ripples into icy silver as snow swirls around it. [8-12s] The snow becomes rain and she is walking across a wide empty rooftop at night, wet and mirror-like, with only soft out-of-focus city lights far away and no buildings close by; the gown is now deep sapphire and reflects in the puddles. [12-15s] The rain clears to a sunrise beach, the gown turns soft gold, she stops, looks into the lens, and the silk billows out behind her. Sound: one continuous elegant electronic beat, desert wind, a snowstorm whoosh, rain on asphalt, gentle waves at the end. Photoreal, editorial lighting. The gown has no logo, no monogram, no text. No on-screen text, no captions, no subtitles.

  • Vertical phone video in a leafy park in autumn. A woman in her thirties walks with a happy golden retriever on a plain orange harness and lead with no logo, filming herself. The dog trots beside her, tail wagging, sniffing leaves. She says to camera: "First walk where he didn't pull once. I'm almost emotional." She laughs. Crunching leaves, distant kids playing. Authentic footage shot on a phone, natural daylight, real skin texture with pores, unretouched, slight handheld movement, realistic ambient sound, no music. Hands stay relaxed; nothing is opened, sprayed, poured or assembled. No logos, no brand names, no readable text anywhere. No on-screen text, no captions, no subtitles.

    Vertical phone video in a leafy park in autumn. A woman in her thirties walks with a happy golden retriever on a plain orange harness and lead with no logo, filming herself. The dog trots beside her, tail wagging, sniffing leaves. She says to camera: "First walk where he didn't pull once. I'm almost emotional." She laughs. Crunching leaves, distant kids playing. Authentic footage shot on a phone, natural daylight, real skin texture with pores, unretouched, slight handheld movement, realistic ambient sound, no music. Hands stay relaxed; nothing is opened, sprayed, poured or assembled. No logos, no brand names, no readable text anywhere. No on-screen text, no captions, no subtitles.

What Minimax H3 is good at

4 to 15 seconds

Any whole number of seconds in one generation.

Up to 2K

A 768P draft first, then the same prompt in 2K, from 21:9 to 9:16.

Lines spoken in the clip

Put a line in quotes and someone says it, with the room sound around it.

First and last frame

Start from a photo and set the frame it lands on.

Up to 15 references

9 images, 3 videos and 3 audio clips keep a product or person the same.

G2

Rated 4.8/5 on G2

We’ve earned 40+ badges and 1000+ reviews on G2.

G2 Best Software 2025 — Top 50 Design ProductsG2 Summer 2025 — Best Results, Small BusinessG2 Summer 2025 — Best Usability, Small Business

Check out other models

Run the same prompt on another model in Model Playground and compare.

Minimax H3 FAQ

Minimax H3, also written Hailuo-03, is a video model from MiniMax’s Hailuo line. On Creatify it makes 4 to 15 second clips at 768P or 2K from a text prompt, a first frame (with an optional last frame), or reference images, videos and audio. Creatify is an independent platform, not MiniMax’s own app.

Make your first Minimax H3 clip

Write a sentence or drop in a photo, and get up to 15 seconds in 2K with sound.