Hailuo H3, explained
Hailuo H3 is the third generation of MiniMax's video model — the same model this site lists as MiniMax H3. Here is what it generates, how long a clip can run, and how to get a first video out of it.
MiniMax H3 (mini-max-h3.com) is an independent third-party tool and is not affiliated with, endorsed by, or operated by MiniMax or Hailuo.
Hailuo, MiniMax and H3 — which name means what
The confusion is understandable: three different names circulate for one model, and search results mix them freely. If you arrived here searching for "Hailuo H3", you are looking at the same generation of video model that other pages call MiniMax H3 — Hailuo is the consumer-facing brand MiniMax ships it under.
One clarification worth stating plainly, because it affects whose documentation you should trust: this site is an independent third-party tool. We call the model by its vendor name so that it is obvious where the model comes from and where the tooling does not.
- Hailuo
- MiniMax's product brand for its video generation tools and apps — the name you see in consumer apps and app stores.
- MiniMax
- The company that builds the model. When a page says "MiniMax H3" it means the same model, named after the vendor rather than the app brand.
- H3
- The generation number. H1, H2 and H3 are successive model releases; H3 is the current one, which is why older benchmarks and prompt examples may not transfer directly.
What Hailuo H3 can generate
The capabilities below are the ones this site exposes through the generator. Treat them as the practical envelope rather than a full model card.
| Input | A text prompt, or a first-frame image plus a prompt. Text-to-video works from an idea; image-to-video keeps a product shot, portrait or logo recognisable while animating it. |
|---|---|
| Resolution | 2K output from either input mode. That is high enough to crop into a vertical edit without a visible quality drop. |
| Clip length | 5, 10 or 15 seconds per generation. Pick the shortest length that covers the shot — regenerating a 5-second clip costs a third of regenerating a 15-second one. |
| Aspect ratio | 16:9 for desktop and ads, 9:16 for short-form vertical, 1:1 for feeds, or matched to your source image. |
| Rendering | Asynchronous. Your request is queued and rendered on the model side, so you can close the tab — the task stays saved and the clip appears when it finishes. |
From nothing to a 2K clip
- 1
Write the prompt, or skip straight to an image
Describe the shot: subject, action, camera movement, lighting and mood. If you already have a still you like, upload it instead and describe only the motion.
- 2
Upload a first frame (optional)
Image-to-video is the reliable path for products and portraits: the first frame pins down identity and composition, so the model only has to invent the movement.
- 3
Choose duration and aspect ratio
5 seconds for a single beat, 15 for a sequence with a camera move. Match the ratio to the destination — vertical for short-form, 16:9 for everything else.
- 4
Generate and review
Generation is queued, so you can leave the page. When it lands, judge the motion first and the polish second — a clean camera move with a slightly wrong detail is usually worth keeping over a beautiful frame that barely moves.
Want to know what a run will cost before you spend it? Use the H3 credit calculator
Prompting is where the quality lives
Most disappointing H3 output is a prompt problem, not a model problem. The guide covers the structure that works — subject, action, camera, light, style — plus中文提示词 examples and the mistakes that produce a static, lifeless clip.
Open the MiniMax H3 prompt guide