Hailuo

Hailuo is MiniMax's AI video app at hailuoai.video; its current model, MiniMax H3 (Jul 31, 2026), makes 4 to 15 second videos with stereo sound at up to 2K and has open weights. H3 ranks 11th on Arena's text-to-video leaderboard as of Nov 6, 2026.

Hailuo is the AI video app of the Chinese AI company MiniMax, at hailuoai.video, and as of Nov 6, 2026 it runs MiniMax H3, a model released on Jul 31, 2026 that generates 4 to 15 seconds of video with native stereo sound at up to 2K, from any mix of text, images, video and audio. MiniMax published H3's weights on Hugging Face under its own community license, making it one of the strongest open video models: 11th on Arena's text-to-video leaderboard with a score of 1456 (Arena, read Nov 6, 2026). Hailuo's subscription prices were not shown on its site without signing in on Nov 6, 2026.

Pick Hailuo or H3 for low-cost 15-second clips with sound, dialogue in 11 languages, or a model you can host yourself on a multi-GPU server. Look elsewhere for the very top of the rankings (Google's Gemini Omni in Google Flow is 1st, Seedance 2.0 6th) or for single takes longer than 15 seconds (Kling 4.0 and Seedance 2.5 do 30). Older Hailuo models, Hailuo 2.3 and Hailuo 02, sit 34th to 38th on the same board. Best AI Video Generator compares them.

Quick facts

Get it hailuoai.video on the web; MiniMax API; weights on Hugging Face
Maker MiniMax
Current model MiniMax H3 (Jul 31, 2026)
Earlier models Hailuo 2.3, Hailuo 02 Pro and Standard (Arena)
Clip length 4 to 15 seconds (Hugging Face)
Output Up to 2K at 24 fps, stereo audio at 32 kHz, aspect ratios from 21:9 to 9:16
Price Subscriptions on hailuoai.video; amounts not shown before sign-in (Nov 6, 2026)
Ranking MiniMax H3 11th on Arena text-to-video, score 1456 (read Nov 6, 2026)
Open weights Yes, MiniMax H3 Community License; users in the USA, EU, UK and South Korea apply through a form

Models

MiniMax H3 treats text, images, video and audio as one context and generates video with sound in the same pass. MiniMax lists multi-shot scenes, editing described in plain language, accurate text and brand logos, and motion transfer from a reference video among its strengths (MiniMax). Its model card adds dialogue in 11 languages (Arabic, Chinese, English, French, German, Italian, Japanese, Korean, Portuguese, Russian and Spanish) and a reference mode that takes up to 9 images, 3 video clips and 3 audio clips, 12 files in all (Hugging Face).

Hailuo 2.3 and Hailuo 02. MiniMax's earlier video models remain on Arena's board: Hailuo 2.3 at 34th (1205), Hailuo 02 Pro 36th and Hailuo 02 Standard 38th (Arena). Hailuo's home page now features only H3. In a Nov 6, 2025 video, Dan Kieft showed Hailuo 2.3 and a faster 2.3 Fast in the Hailuo app, with start and end frames, preset motions and camera controls, and clips of up to 6 seconds at 1080p or 10 seconds at 768p at the time.

Price

MiniMax has not published H3's per-second price on its blog; it says only that at 2K H3 costs less than a third of what mainstream models charge per second, and at 768p less than half their 720p price (MiniMax, Jul 31, 2026). That is the maker's own comparison. H3 is also sold inside Luma and Pika, whose Pika H3 Flex is a further-trained H3.

Running H3 yourself

The released H3 is a 33-billion-parameter model in BF16, of which about 13 billion sit in branches that inference does not need to load; MiniMax's own serving example uses four GPUs (Hugging Face). Two pieces stay closed: the H3-Context-IR preprocessing system, available only through the API, and the 2K upscaling module, promised later. So the open weights are for servers, not a home PC; for a model that runs on one consumer graphics card, see Wan and Run AI Locally.

Features

Hailuo Agent and templates. Hailuo's app offers video, image and audio creation, a Hailuo Agent, an Omni Reference input and ready-made templates for ads and short effects (Hailuo).

Company. MiniMax says its models and AI products serve more than 300 million users in over 230 countries and regions, and more than two million businesses and developers (MiniMax investor relations). It also makes the MiniMax M3 language models and Speech and Music models.

Hailuo vs the alternatives

Google Flow with Gemini Omni 1.2 Flash ranks 1st on Arena (1541 to H3's 1456) and is free for 50 credits a day; pick it for quality, H3 for open weights.

Seedance 2.0 and 2.5 rank 6th and 7th and make up to 30 seconds; neither has open weights.

Kling 4.0 makes 30-second takes but has no public ranking yet.

Wan is Alibaba's open model family; Wan 2.2 runs on a single 24 GB card, though it ranks far lower (44th).

Pika sells Pika H3 Flex, built on H3, from $10 a month.

Frequently asked questions

What is Hailuo AI? Hailuo AI is MiniMax's video generation app at hailuoai.video; as of Nov 6, 2026 it runs MiniMax H3, which makes up to 15 seconds of video with stereo sound.

Is MiniMax H3 open source? MiniMax H3's weights are public on Hugging Face under the MiniMax H3 Community License, not a standard open-source license, and users in the USA, EU, UK and South Korea must apply through a form.

How long are Hailuo videos? MiniMax H3 in Hailuo makes videos of 4 to 15 seconds at up to 2K resolution.

How does Hailuo rank against Veo and Kling? On Arena's text-to-video leaderboard read Nov 6, 2026, MiniMax H3 is 11th, above Veo 3.1 (14th) and Kling 2.6 Pro (31st) but below Gemini Omni (1st) and Seedance 2.0 (6th).

How much does Hailuo cost? Hailuo sells monthly subscriptions, but the amounts were not shown on hailuoai.video without signing in on Nov 6, 2026; MiniMax says H3 costs less than a third of mainstream models per second at 2K.

Sources

Moments

NEW Hailuo 2.3 AI Video Generator is INSANE for Cinematic AI Videos

Dan Kieft

  1. Hailuo 2.3 and Hailuo 02. MiniMax's earlier video models remain on Arena's board: Hailuo 2.3 at 34th (1205), Hailuo 02…

See also

About this page

Last updated Nov 6, 2026 by the Vi.ki team. Page text CC BY-SA 4.0.

Edit History

Hotkeys

/
Jump to the search box
? or Ctrl + /
Open this list
Esc
Close this list, the photo viewer, or the account menu

Video

Space or K
Play or pause
← →
Back or forward 5 seconds
J L
Back or forward 10 seconds
0 to 9
Go to 0% to 90% of the video
, .
Previous or next frame, while paused
M
Mute or unmute
C
Captions on or off
F
Full screen on or off
A D
Previous or next moment

Photo viewer

← → or A D
Previous or next photo
↑ ↓ or W S
Previous or next photo
Home End
First or last photo
I
Show or hide the photo's info
Z
Clickless on or off: a pointer over a thumbnail shows it
N
Compare with the last photo: back to the one shown before this, and N again comes back
Q
Move the thumbnails to the next edge: top, bottom, left and round
, or Ctrl + ,
Open or close the viewer's settings
V
Clickless pan: a zoomed photo follows the pointer, rather than being dragged
Shift
Browse on or off
B
Show or hide the button bar
T
Show or hide the thumbnails
1
100% zoom on or off
0
Rate the photo as no opinion; again takes the answer back
+ -
Rate the photo a step higher or lower
F
Full screen: the photo alone. F again puts everything back.
Back to top