Sand.ai
Founded by Marr Prize winner Cao Yue, Sand.ai specializes in video generation technology and products. On the model front, the company has independently developed Magi-1 (an autoregressive video model) and GAGA-1 (a model for simultaneous audio-visual generation), and recently released MAGI-2 Preview—the world's first 100-billion-parameter MoE video generation model—as a fully open-source project. On the application front, VidMuse pioneered an audio-driven paradigm that uses musical structures to orchestrate visuals, achieving an Annual Recurring Revenue (ARR) of over $10 million within just two months of launch. This dual-engine approach accelerates the mission of making intelligence universally accessible to humanity.
Core Mission
Accelerating AI to Benefit Everyone
Create the Future
- Leverage video generation models to enable enthusiasts to create video content with low barriers to entry, low costs, and short turnaround times.
- Reduce the marginal cost of visual content creation by five orders of magnitude.
- Explore optimal solutions for video compression.
Founder of Sand.ai
- Co-founder of Lightyear (Qian Guang Nian Zhi Wai); Head of a research center at the Beijing Academy of Artificial Intelligence (BAAI); formerly a Principal Researcher at Microsoft Research Asia.
- Represents world-class technical expertise; author of Swin Transformer; recipient of the ICCV 2021 Best Paper Award (Marr Prize—one of the highest honors in computer vision); over 40,000 citations for a single paper and over 80,000 total citations; Microsoft Research Fellow; recipient of the Tsinghua University Top-Grade Scholarship.
- Holds a Bachelor's degree in Computer Software and a Ph.D. in Software Engineering from Tsinghua University.
