MiniMax: The Chinese AI Lab Doing More With Less

MiniMax pursues Artificial General Intelligence (AGI) by prioritizing cost-efficient, multimodal, and open-source development, challenging the capital-intensive, closed-system approach common in Silicon Valley. By focusing on inference efficiency, the company delivers 95% of frontier model performance at a tenth of the cost, enabling broader access to advanced intelligence. Integrating language and video models into a unified system allows for complex, agentic tasks, while an open-source strategy—exemplified by the H3 model—fosters rapid innovation through developer feedback loops. With annualized revenue reaching $800 million, largely driven by B2B enterprise adoption, MiniMax demonstrates that resource constraints can drive superior engineering. CEO Dr. Yan Junjie emphasizes that AGI must be a shared global resource, positioning the company to scale intelligence through optimized post-training rather than relying solely on brute-force compute.