MiniMax Hailuo H3 Max Text to Video

MiniMax H3 Max Text to Video

What is MiniMax H3 Max?

MiniMax H3 Max is the top-quality tier of MiniMax's H3 (Hailuo-03) video model, a general-purpose multimodal system that turns a single text prompt into a finished video clip. Where the lighter H3 tiers trade fidelity for speed, H3 Max is tuned for the hardest prompts — scenes packed with multiple subjects, fast action, fluids, smoke and reflective materials — and keeps them coherent instead of smearing. Every clip is generated with native stereo audio in the same pass, so motion and sound arrive together.

Key Features

• •Text-to-video from a prompt up to 7,000 characters
• •Selectable output resolution: 480P for quick drafts, 768P for finals
• •Clip length from 5 to 15 seconds
• •Aspect ratios for every screen: 21:9, 16:9, 4:3, 1:1, 3:4 and 9:16
• •Native stereo audio generated alongside the picture
• •Prompt-expansion modes (disabled, balanced, quality) to control how much the prompt is enriched

Best Use Cases

H3 Max is the tier to reach for when a shot is demanding. It suits cinematic teasers and establishing shots, product and science demos, sports and dance motion, explainer beats, and social clips where one legible action has to read at a glance. In testing, a cryogenics-lab prompt produced dense volumetric fog that billowed over a bench and rolled across the floor while glass beakers stayed sharp and reflective — the kind of fluid physics that trips up lighter models.

Prompt Tips and Output Quality

Lead with one clear primary action, name concrete subjects and materials, and let the camera move follow that action. Keep prompt expansion on balanced for a well-written prompt; switch to quality to enrich a sparse one. Choose 768P and 16:9 for finished landscape shots, or 9:16 for vertical social. At 768P the output is clean and physically plausible, with motion and native audio in sync.

How to Use MiniMax H3 Max Text to Video

MiniMax H3 Max turns a text prompt into a video with native audio at up to 768P. It is the highest-quality H3 tier, so it rewards prompts that describe a specific scene and one clear motion.

Writing the prompt

Open with the subject and the single most important action, then add the setting, materials and a camera move motivated by that action. Name concrete things — a gloved researcher tipping a steel flask, glass beakers, cold blue light — rather than stacking adjectives. One primary action reads far more cleanly than several competing ones. The prompt field accepts up to 7,000 characters, but tight, concrete prompts outperform long ones.

Choosing settings

• •Resolution: 768P for anything you intend to publish; 480P for quick drafts and iteration.
• •Duration: 5 to 15 seconds. Six to eight seconds is enough for a single action to play out; reserve longer clips for a sequence of beats.
• •Aspect ratio: 16:9 for landscape, 9:16 for vertical social, 1:1 for square feeds.
• •Prompt expansion: balanced suits a well-written prompt; quality enriches a sparse one; disabled keeps your wording verbatim.

Recommended Settings

For a polished default, use 768P, 16:9, duration 6 and balanced prompt expansion — the configuration behind the model's default example. That cryogenics-lab clip rendered dense, volumetric fog and reflective glassware cleanly, with motion and native audio in sync and no visible artifacts. Drop to 480P while you are testing phrasing, then re-run your final at 768P. If a scene has several moving elements, simplify to the one that matters most, confirm it renders well, and add complexity only once the core shot holds together.

Examples to try

Each example below ran end to end through this model at the settings shown on its card; clicking one loads its prompt into the form. Watch how the length, motion, framing and native stereo audio all carry straight from the prompt.

• •Demanding fluid physics (the default). Prompt: "Inside a cryogenics lab, a gloved researcher tips a steel flask of liquid nitrogen across a stainless bench; dense white fog cascades over the edge and the camera drops low to follow it rolling across the floor, swallowing a row of glass beakers as frost blooms up the flask under cold blue light." The fog billows and rolls while the glassware stays sharp and reflective. Keep one primary action with a motivated camera move, and render finals at 768P.
• •Native stereo audio. Prompt: "A knob of butter drops into a screaming-hot cast-iron skillet and bursts into a sizzle; a hand tosses in sliced shiitake mushrooms that crackle and brown as steam curls up over a blue gas flame. Sound: butter sizzling and mushrooms crackling in the pan." The clip plays back with a synced sizzle over the bubbling butter. Name the sound in a short "Sound:" clause to steer the audio; 480P is enough when the point is the soundtrack.
• •Fast action without smearing. Prompt: "A blue rally car power-slides through a gravel forest hairpin, its back end stepping out as a spray of stones and dust fans off the rear tyres while a chase camera tracks alongside through the drift." The car stays coherent through the drift as stones and dust fan off the tyres. Lead with the single action and let the chase camera follow it — fast motion reads even at 480P.

More to try: "A barista steams milk as the wand hisses and the pitcher fills", "Two fencers lunge and parry across a piste, blades flashing", "Sparks shower off an angle grinder cutting steel on a workbench".

Does MiniMax H3 Max generate sound?

Yes — every clip comes back with native stereo audio made in the same pass as the picture.

What resolutions can it output?

480P or 768P; use 768P for final quality and 480P for fast drafts.

How long can a clip be?

From 5 to 15 seconds, set with the duration control.

Does it need an input image?

No. This is text-to-video; for first and last-frame animation use the MiniMax H3 Max image-to-video model.

What makes the Max tier different?

It is the highest-quality H3 tier, built to hold complex, physically demanding prompts together where lighter tiers drift.