Langfuse

Langfuse is an open-source LLM engineering platform that helps developers trace, monitor, evaluate, and improve AI applications and agents at production scale. They enable teams to analyze LLM interactions, manage prompts, evaluate outputs, and monitor cost and latency, making it easier to build reliable AI systems with continuous optimization. The platform supports observability, experimentation, and prompt management in one connected workflow, allowing developers to move from prototype to production efficiently.

They provide a comprehensive toolkit with hierarchical tracing, human annotation workflows, evaluation systems, and real-time analytics dashboards, enabling teams to debug and improve AI applications using production data. Developers can integrate with OpenAI, Anthropic, LangChain, CrewAI, Gemini, and 80+ additional frameworks and tools, while benefiting from OpenTelemetry-native support, scalable infrastructure, and self-hosting options. With MIT-licensed open-source architecture, enterprise-grade security, and support for billions of monthly observations, they are designed for AI engineering teams building large-scale LLM applications.

And the best part? They offer a free tier with 50,000 observations per month, while advanced and enterprise pricing is not publicly disclosed.