China's MiniMax M2 Ranks Fifth on AI Index Beating Major Rivals

Chinese AI startup MiniMax has released its open-source large language model, MiniMax M2, which ranks fifth on the Artificial Analysis Intelligence Index, outperforming models such as Google’s Gemini 2.5 Pro and Anthropic’s Claude Sonnet 4.5, particularly excelling in coding and agentic reasoning tasks. Despite having 230 billion parameters, MiniMax M2 activates only 10 billion at a time, allowing near frontier-level performance with high efficiency, running at about 8% of the cost of competing models and nearly twice as fast. The model supports end-to-end developer workflows and long-horizon toolchains, making it well-suited for autonomous agents and multi-modal intelligence applications. Backed by major investors like Alibaba, Tencent, and Hillhouse, MiniMax aims to democratize AI access through open-source releases and affordable pricing, potentially reshaping global AI adoption. MiniMax M2's code generation and tool-use capabilities place it among the strongest open-source AI models, reinforcing China's growing presence in advanced AI development. The company has made the model weights publicly available on platforms like Hugging Face and GitHub, encouraging broad deployment and innovation.