LangfuseAn open-source LLM engineering platform offering observability, metrics, evaluations, and prompt management to debug and enhance large language model applica...

Langfuse is an open-source LLM (Large Language Model) engineering platform designed to help developers build, debug, and enhance AI agents and applications. It provides a comprehensive set of tools for observability, metrics, evaluations, and prompt management, allowing teams to collaborate and continuously improve the quality, cost, and latency of their AI products. Langfuse supports the entire LLM engineering loop, from prototype to production, by integrating observability, prompt management, evaluations, experiments, and human annotation into a single workflow. The platform offers features such as hierarchical tracing, which captures every LLM call, tool invocation, and retrieval step, and allows filtering by user, session, cost, latency, or custom metadata. It also supports various evaluation methods, including LLM-as-a-judge, heuristic functions, and human review. Additionally, Langfuse provides a prompt management system with one-click deployments and rollbacks, a playground for testing prompts, and experiment definition and comparison tools. Langfuse works with any language and framework that supports OTel instrumentation and has over 100 integrations with popular agent frameworks, model providers, and other tools. It is an open platform with an MIT license, allowing users to self-host at scale and contribute to the community. Langfuse is used by over 100,000 engineers and 19 of the Fortune 50 companies, handling over 10 billion observations per month.