All NewsEducationTV
Equities & FundsCrypto & Digital AssetsAI & TechnologyBusiness & CorporateUS Politics & PolicyGeopolitics & Global RiskMacro, Rates & FXCommodities & EnergyEuropean Politics & MarketsAsia-PacificReal Estate & Property
Story archiveAll categories
← All Stories

China's MiniMax releases H3 video model

Created at 31 Jul · 4:45 AM1 source↑ Market-relevant
IN SHORT

Chinese AI firm MiniMax has launched its H3 video-generation model, capable of producing up to 15-second clips in 2K resolution with stereo sound. The model supports text, image, video, and audio inputs and can edit existing content. MiniMax plans to release model weights for customization, extending the open-weight approach to video generation.

✉Newsletter

PiQ Daily

Pick your topics. Get only what matters, on your cadence.

Key Numbers

15 secondsmaximum video length
2K resolutionvideo output resolution
2022MiniMax founding year
2.7 trillionparameter language model under development
one-thirdcost reduction compared to rivals

Who's Involved

MiniMax
Chinese AI firm that released the H3 video model
ByteDance
Rival AI company that released Seedance 2.0
Kuaishou
Rival AI company that released Kling 3.0
Eduardo Baptista
Reuters reporter
Stephen Coates
Reuters editor

↳ Why This Matters

MiniMax's H3 model release signifies increased competition in the AI video generation market, particularly with its open-weight approach and focus on commercial applications. The model's design to work with Chinese chips also aligns with China's broader strategy to reduce dependence on U.S. technology.

Key facts

  • Chinese AI firm MiniMax has launched its H3 video-generation model.
  • The H3 model can generate videos up to 15 seconds long in 2K resolution with native stereo sound.
  • It can process text, images, video, and audio inputs, and edit existing content.
  • MiniMax plans to release the model's weights for users to download and customize.
  • The model is designed for commercial applications such as advertising, e-commerce, product design, and games.
  • H3 is engineered to operate with Chinese-made semiconductors.

Chinese AI firm MiniMax has launched its new H3 video-generation model, intensifying competition in the rapidly expanding market. The model, capable of producing up to 15-second videos in 2K resolution with stereo sound, can process and edit content using text, image, video, and audio inputs. MiniMax intends to release the model's weights, allowing users to download and customize the system, a move that extends the open-weight approach to video generation.

Founded in 2022, MiniMax is a well-funded startup competing with domestic tech giants and global leaders in the AI foundation model space. The company, which went public in Hong Kong in January, aims to target commercial applications such as advertising, e-commerce, product design, and gaming with its H3 model, claiming it will cost less than a third of comparable rival products. The development comes amid a broader push in China's AI sector to reduce reliance on U.S. semiconductors, with H3 designed to function with Chinese-made chips.

Frequently asked questions

H3 is a new video-generation model developed by Chinese AI firm MiniMax that can create up to 15-second videos in 2K resolution with stereo sound, processing various input formats like text, images, video, and audio.

The H3 model supports text, image, video, and audio inputs, can edit existing content, transfer movements between videos, and will be released with open weights for customization.

Competition is intensifying, with rivals like ByteDance (Seedance 2.0) and Kuaishou (Kling 3.0) also releasing advanced video generation models.

MiniMax has designed the H3 model to be compatible with Chinese-made chips, supporting China's goal of reducing reliance on U.S. semiconductors.

What Happens Next

01MiniMax plans to release H3's model weights within days.

Get the newsletter.

Pick the topics you actually care about. We'll email when there's news worth your time, on the cadence you choose. Cancel any time from your account.

Cadence

How It Developed

MiniMax released its H3 video-generation model.
The H3 model can generate 15-second videos in 2K resolution with stereo sound.
The model supports text, image, video, and audio inputs.
MiniMax plans to release H3's model weights for customization.
The company aims H3 at commercial applications like advertising and e-commerce.
H3 is designed to work with Chinese-made chips.

Sources

T1
China's MiniMax releases H3 video modelReuters

Related Stories

OpenAI cuts prices on smaller models amid business cost scrutiny
30 Jul · 5:03 PM
Unitree Robotics launches $620 million IPO on Shanghai's STAR Market
30 Jul · 4:31 PM
ByteDance projects $4B AI revenue, restructures Lark office tool
30 Jul · 12:36 PM
Microsoft considers open-weight AI amid rise of Chinese models
30 Jul · 12:41 PM
BYD to debut humanoid robots in August
30 Jul · 8:06 AM