langfuse/langfuse

Health v2 · maintenance only

Flags: no_license

Adoption not part of the score

33767 stars · 3650 forks observed · 2026-08-28

What it is AI-extracted, prompt v1, taxonomy v1, 2026-08-29, confidence not recorded

Langfuse is an open-source AI engineering platform for LLM observability, tracing, evaluation, prompt management, and metrics, deployable as a cloud service or self-hosted. It integrates with OpenTelemetry, LangChain, OpenAI SDK, LiteLLM, and 100+ frameworks via Python/JS SDKs to trace, evaluate, and improve AI agent applications.

Use cases

• trace and debug LLM calls and agent steps in production
• evaluate LLM app output quality with LLM-as-judge and human review
• manage and version prompts separately from code
• monitor LLM cost, latency, and quality with dashboards and alerts
• run experiments comparing prompts and models on test datasets
• self-host an LLM observability platform
• collect user feedback and human annotations on traces

When to choose

• you need production observability and tracing for LLM or agent applications
• you want integrated prompt management, evals, and metrics in one platform
• you need OpenTelemetry-based tracing with broad framework integrations
• you want a self-hostable alternative to closed LLM monitoring tools

When to avoid

• you only need simple application logging without LLM-specific features
• you need a lightweight library-only solution with no server component
• you require a fully permissive license for commercial redistribution without checking terms

Facets

service · maturity active

monitoring tracing analytics prompt-engineering llm-inference alerting data-visualization developer-tools large-language-models monitoring analytics developer-tools self-hosted self-hosted cloud python llm-observability llm-evaluation prompt-management llmops opentelemetry agent-evals llm-as-judge tracing playground human-annotation ai-agents docker web-server nodejs typescript

Member repositories

For agents

markdown · JSON · MCP: product_card(name="langfuse/langfuse")