Xuhui-based AI company unveils groundbreaking products

Shanghai-based AI unicorn MiniMax recently hosted a highly anticipated tech launch week event, introducing five revolutionary products that push the boundaries of base models, multimodal technology, and intelligent agents.

The star attraction, MiniMax-M1, is the world's first open-source large-scale hybrid architecture inference model. It delivers top-tier performance in productivity scenarios, rivaling leading closed-source models while significantly reducing computational costs.

Another highlight, the Hailuo-02 video generation model, achieves world-class quality and cost efficiency, challenging international benchmarks. Its innovative architecture enables complex physical scene rendering, setting a new standard for video AI.

MiniMax also introduced the MiniMax Agent, a versatile intelligent agent capable of executing complex, multi-step tasks, and the Hailuo Video Agent, China's first "one-click" video production assistant. Additionally, Voice Design, an upgrade to the Speech 02 model, allows for highly customizable voice synthesis.

These innovations underscore MiniMax's commitment to autonomous innovation, breaking through computational barriers and fostering global AI accessibility. As a leader in Xuhui district's AI ecosystem, MiniMax joins six other industry giants to form a powerful "Big Dipper" matrix, driving technological advancement and industry collaboration in Shanghai.

The MiniMax display panel at the World Artificial Intelligence Conference in Shanghai on July 5, 2024. [Photo/IC]