Galileo

Galileo is an AI observability and evaluation platform for GenAI applications and agents. It offers offline evaluation with 20+ built-in metrics, production monitoring with custom dashboards, and real-time guardrails powered by proprietary Luna small language models. Luna models replace expensive LLM-as-judge evaluators with sub-200ms latency and 97% cost reduction. Integrates with CrewAI, LangGraph, OpenAI Agent SDK, LlamaIndex, and Amazon Strands. Also publishes the Hallucination Index, a public benchmark ranking LLMs by hallucination rates.