Soon

Hailuo H3 AI Video Generator

Create cinematic AI videos with the Hailuo family. MiniMax's frontier multimodal video model H3 is expected in Q3 2026 — generate with Hailuo 2.3 today: 6–10 second clips at up to native 1080p, from text or a start image.

Hailuo H3 is coming soon — prompts run on Hailuo 2.3 in the meantime.

How It Works

How to Generate Videos with Hailuo

From first prompt to production-ready video in four steps.

Step 1

Pick a Hailuo model

Choose a Hailuo model from the model selector in the video generator.

Step 2

Add a start frame

Upload a still image to animate with image-to-video, or start from text alone.

Step 3

Write the prompt

Describe the shot like a director: subject, camera, lighting, and mood.

Step 4

Generate and refine

Compare takes, then enhance and upscale the best one — all in one place.

Hailuo H3 Technical Specifications

Frontier

Multimodal video generation — MiniMax’s next flagship model

Q3 2026

Expected release window

2.7T

Parameters in M3 Pro, the LLM launching alongside H3

1080p

Native output resolution on Hailuo 2.3 today

6–10s

Clip durations on current Hailuo models

Start frame

Image-to-video from a single still

Expressive

Character micro-expressions held across the full clip

What Hailuo H3 Can Do

01 Dynamic motion

Dynamic Motion That Feels Physical

The Hailuo family is known for realistic physical motion, stable subjects, and expressive camera work. Hailuo 2.3 raised the bar for physical actions, stylization, and motion-command response — the baseline H3 is built to surpass.

Open the generator

02 Text-to-video

Cinematic Text-to-Video

Describe a shot like a director — subject, camera, lighting, mood — and get a cinematic clip. Current Hailuo models generate 6 or 10 second videos at up to native 1080p from a text prompt, the workflow H3 slots straight into.

Open the generator

03 Characters

Characters and Micro-Expressions

Hailuo models excel at human performance: facial micro-expressions, natural gestures, and consistent characters across a shot. Animate a still with image-to-video, or direct a scene entirely from text.

Open the generator

Prompt Examples

A grassy surface, clear skies, rocky desert, millions of red poppy flowers slowly grow out of the grass.

Try this prompt

Hailuo Prompt Gallery

Copy our curated prompts and start creating cinematic AI videos right now.

GoPro footage of someone airgliding through the mountains, shaky camera footage.

A ship sailing through rough waters, grey skies, cinematic ocean spray.

A dragon flying over New York City, drone shot.

Asteroids and diamonds and stars floating in the skies, 4k high quality, camera does not move.

A silver orb, abstract translucent shape, is floating through the streets of London, shaky camera footage.

A grassy surface, clear skies, rocky desert, millions of red poppy flowers slowly grow out of the grass.

The woman looks shocked and surprised, camera slowly zooms out.

GoPro footage of someone airgliding through the mountains, shaky camera footage.

Browse all prompts

Frequently Asked Questions

What is Hailuo H3?

Hailuo H3 is MiniMax’s upcoming frontier-level multimodal video generation model — the next generation of the Hailuo AI video family after Hailuo 02 and Hailuo 2.3. It was reported by Reuters on July 8, 2026 as part of MiniMax’s next model launch.

When is the Hailuo H3 release date?

MiniMax has not announced an exact date. Reporting indicates the new model family could launch as early as Q3 2026. This page is updated as soon as MiniMax confirms details.

Does Hailuo H3 have 2.7 trillion parameters?

Not exactly. The 2.7-trillion-parameter model in the announcement is M3 Pro, a large language model launching alongside H3 — about six times the size of MiniMax’s current flagship and the largest model announced by a Chinese lab. Hailuo H3 is the video generation model in the same family; MiniMax has not published its parameter count.

Is Hailuo H3 available yet?

Hailuo H3 is not released yet, so it is not available anywhere. MiniMax’s current Hailuo models — Hailuo 2.3, Hailuo 2.3 Fast, and Hailuo 02 — are available today, and prompts from the generator on this page run on Hailuo 2.3 in the meantime.

How is Hailuo H3 different from Hailuo 2.3?

Hailuo 2.3 is MiniMax’s current released video model, known for realistic motion, stylization, and character micro-expressions at up to native 1080p. H3 is described as a frontier-level multimodal model, which suggests a step up in scale and capability, but MiniMax has not published specs or benchmarks yet.

Will Hailuo H3 be open source?

MiniMax plans to open-source the 2.7T-parameter M3 Pro language model, which would make it the largest open-source model from a Chinese firm. Whether H3 itself will be open-sourced has not been confirmed.

How do I try Hailuo AI video generation right now?

Type a prompt in the generator at the top of this page and click Generate — your prompt is loaded into a video tool running the current Hailuo models. Clips are 6 or 10 seconds at up to 1080p, with text-to-video and image-to-video support.

Hailuo H3 by MiniMax

MiniMax's frontier multimodal video model, expected as early as Q3 2026. Until then, Hailuo 2.3 and Hailuo 02 are available now — generate cinematic clips today.

Try Hailuo Now

Page last updated July 15, 2026