- Observability: Langfuse

Open source observability and analytics for LLM applications, providing production tracing, prompt management, and automated evaluation.

Langfuse delivers a dedicated engineering stack for LLM operations (LLMOps). It tracks nested traces across complex chains, captures precise token usage for cost monitoring, and manages prompt versioning via a central registry. Teams use its SDKs for Python and JS to instrument RAG pipelines, enabling them to run model-based evaluations (LLM-as-a-judge) and collect human feedback directly from production. The platform integrates with tools like LiteLLM and LangChain to surface latency bottlenecks and quality regressions across thousands of generations.

https://langfuse.com