Langfuse
Open-source LLM observability platform for tracing, evaluating, and iterating on AI application performance in production.
About
Langfuse gives engineering teams a complete toolkit for understanding and improving language model applications. Its tracing layer records every LLM call, embedding operation, and agent step so developers can pinpoint where things go wrong in complex workflows. Prompt management lets teams version, deploy, and test prompts across environments using a built-in playground. Evaluation tools cover LLM-as-judge scoring, custom code evaluators, user feedback collection, and dataset-based regression testing. Langfuse is open-source and self-hostable, making it accessible at no cost for teams that prefer to run their own infrastructure, with a managed cloud option also available.