Luna 2, the next generation of small language models, is designed for customized real-time guardrailing in complex multi-agent systems. It offers low-latency evaluations, cost-effectiveness, and adaptability, allowing organizations to leverage both custom fine-tuned LLMs and SLMs as judges. Luna 2 eliminates the tradeoff between comprehensive evaluation and efficient agents, enabling production-grade AI systems at scale. With its proprietary optimized inference engine and modern GPU hardware, Luna 2 delivers enterprise-grade evaluation with superior out-of-the-box metrics for agentic evaluation and reliability. It also offers customization at speed, allowing enterprises to fine-tune their models in minutes, not months. Available now, Luna 2 powers evaluation across the Galileo platform, making comprehensive agent evaluation and real-time guardrails economically viable.