langfuse

Langfuse Python SDK - LLM observability/tracing, datasets, experiments, LLM-as-a-judge evaluation, and prompt management

langfuse Permissive license MIT Active v4.14.4 released 2026-08-11

Install

langfuse on PyPI

pip

pip install langfuse

uv

uv add langfuse

poetry

poetry add langfuse

Package facts

License	MIT (permissive)
Python support	supports the current Python release (<4.0,>=3.10)
Install friction	low — pure-Python wheel
Runtime dependencies	8 — httpx, pydantic, backoff, wrapt, packaging, opentelemetry-api, opentelemetry-sdk, opentelemetry-exporter-otlp-proto-http
Maintenance	actively maintained — 2 days since the last release
First released	2023-07-12
Popularity	one of the top 1,000 most-downloaded packages on PyPI (30-day window, as of 2026-08-13)
Known vulnerabilities	none known (OSV.dev, checked 2026-08-13)

Evidence: langfuse-4.14.4-py3-none-any.whl

About langfuse

from the package's own PyPI description — quoted content, verbatim

Langfuse Python SDK

MIT License (image) CI test status (image) PyPI Version (image) GitHub Repo stars (image) Discord (image) YC W23 (image)

The Langfuse Python SDK covers the full...

Read as markdown · JSON record

AI interpretation — verify before relying

AI-generated interpretation of the package facts above; every digit, version, license, or vulnerability id it cites is grounded in the facts already shown on this page

Langfuse is a Python SDK for LLM observability, tracing, and evaluation that provides OpenTelemetry-based instrumentation, dataset management, experiment tracking, and prompt management capabilities.

Low friction installation with a pure-Python wheel. Active maintenance (released 2 days ago) and top-1000 PyPI popularity. Eight runtime dependencies are all well-established (httpx, pydantic, opentelemetry-api/sdk) with no known vulnerabilities.

MIT license permits commercial use, modification, and distribution with minimal restrictions—standard permissive terms suitable for most projects.

Usage

pip install langfuse

from langfuse import get_client

langfuse = get_client()
with langfuse.start_as_current_observation(as_type="span", name="my-span") as span:
    span.update(output="result")
langfuse.flush()

Requires Python 3.10 or later (supports current versions, <4.0). Environment variables LANGFUSE_PUBLIC_KEY, LANGFUSE_SECRET_KEY, and LANGFUSE_BASE_URL must be set.

Verdict: Langfuse is a mature, actively maintained observability SDK for LLM applications with low install friction, no known vulnerabilities, and permissive MIT licensing. It integrates OpenTelemetry standards and supports OpenAI/LangChain workflows. Suitable for production use in LLM tracing and evaluation pipelines.

Needs verification

Whether the v4 rewrite (March 2026) introduced breaking changes affecting existing integrations beyond the documented migration guide
Performance overhead and latency impact of tracing on typical LLM application response times
Data retention and privacy guarantees for traces sent to Langfuse platform

llm observability tracingopentelemetry python sdklangchain integration monitoringllm evaluation and datasetsprompt management platformai experiment trackingllm-as-a-judge evaluationopenai tracing integration