Founded two and a half years ago, MiniMax is sharing its mission and vision with everyone for the first time. The MiniMax large model has achieved an average of 3 billion AI interactions per day, processing 300 trillion tokens daily. The application of next-generation model technology based on the MOE (Mixture of Experts) architecture and Linear Attention, along with several latest research developments, was also presented during the partner day.
MiniMax leverages technology to achieve the goal of “3 billion interactions with the world daily,” with the vision of "Intelligence with Everyone." We believe this is the most efficient and the only way to realize AGI. During this partner day, MiniMax announced the release of the video model abab-video-1, the music model abab-music-1, and updated the speech model abab-speech-1. Additionally, MiniMax will launch the multimodal model abab 7 using MoE + Linear Attention technology in the coming weeks.
MiniMax founder Yan Junjie delivered a keynote presentation, highlighting more exciting features that are presented in this article.
MiniMax is the leading large model company in China in terms of daily processing volume and interaction duration.
- The MiniMax large model interacts with global users 3 billion times daily, including
- Processing over 300 trillion text tokens each day, equivalent to experiencing 3,000 lifetimes in one day;
- Generating 20 million images daily, comparable to the art collection of 400 Forbidden Cities;
- Synthesizing 70,000 hours of voice each day, which is equivalent to reading 7,000 books in a single day.
Intelligence with Everyone
Many times, it is not our technology that helps users; rather, it is the users who help us.
Our users are diverse, creative, and full of energy. It is their participation and feedback that lead to better intelligence.
Innovation and breakthroughs in technology stem from people's faith in and commitment to technology. MiniMax will continue to drive the accelerated development of AI through technological advancements, expanding the boundaries of human wisdom.
Currently, both the MiniMax video model and music model are available on the open platform and the Hai Luo AI web version. In the future, we hope to collaborate with more ecosystem partners to explore the practical applications of music and video models, empowering creation and imagination through technology.
Together with all of you, we strive to truly achieve "Intelligence with Everyone."
