OpenViSTAI · Agents · Blockchain · Crypto
Submit a project
Menu
AI Evaluation & Observability

Langfuse

Langfuse is an open-source LLM engineering application that helps teams develop, monitor, evaluate, and debug AI applications by collecting traces of model calls and related application logic. It also provides prompt management, evaluation workflows, datasets, and a playground, and can be used through Langfuse Cloud or self-hosted.

Explore the practical guide ↓

Independent OpenViST project profile, based on linked sources.

Langfuse — OpenViST project profile in AI Evaluation & Observability.
OpenViST project illustration. See the linked sources for product details.
Understand the project

Practical overview

When it fits

Use Langfuse when you need a shared engineering workspace to follow LLM application behavior through traces and connect that observability to prompt iteration and evaluation.

Who it is for

  • Developers instrumenting Python LLM applications.
  • AI application teams investigating model calls, retrieval, embeddings, or agent actions.
  • Teams managing prompts, evaluation datasets, and feedback workflows.

Documented uses

  • Trace model calls and related application logic.
  • Inspect logs and user sessions to debug complex LLM workflows.
  • Manage and collaboratively iterate on versioned prompts.
  • Evaluate applications with judges, code evaluators, feedback, manual labels, or custom pipelines.
  • Use datasets for benchmarks, experiments, and pre-deployment testing.
Before you choose

Requirements & limits

The project documents Langfuse Cloud and self-hosted deployment. Its quickstart requires creating an account or self-hosting, then creating a project and API credentials for the trace-ingestion workflow.

  • Connecting an application requires instrumentation; the documented Python quickstart uses the Langfuse SDK and OpenAI integration.
  • Evaluation options include LLM-as-a-judge, code evaluators, user feedback, manual labeling, and custom pipelines.
  • Datasets support benchmarks, experiments, and pre-deployment testing.
  • Self-hosting is available alongside the hosted Cloud option.
  • The README identifies the project as open source and displays an MIT license badge.

Check current requirements in the official documentation. Source-based guidance does not mean OpenViST has run or independently tested this project.

Source-based practical guide

Trace a Python OpenAI call with Langfuse

Set up a Langfuse project and instrument a Python OpenAI call so its trace can be viewed in Langfuse.

Before you start

  • A Langfuse account or a self-hosted Langfuse instance.
  • Python application using the OpenAI integration.
  • A new Langfuse project and API credentials created in its project settings.

Steps

  1. Create a Langfuse account or self-host Langfuse, then create a project and generate API credentials in its settings.

  2. Install the Python packages with the command below.

    Documented command
    pip install langfuse openai

  3. Set LANGFUSE_SECRET_KEY and LANGFUSE_PUBLIC_KEY in a .env file, and set LANGFUSE_BASE_URL to the appropriate Langfuse Cloud region URL shown in the quickstart.

  4. Use @observe() on Python functions and call the OpenAI integration through from langfuse.openai import openai; run the application code.

  5. View the resulting language-model calls and application logic as traces in Langfuse.

Expected outcome

The instrumented OpenAI call and observed application functions are logged as traces that can be inspected in Langfuse.

Guide limitations

  • The shown code path is for Python and the Langfuse OpenAI integration; other models and frameworks use other documented integrations.
  • The quickstart shows Langfuse Cloud region URLs; it does not specify the base URL for a self-hosted installation.
Documentation 1 ↗Documentation 2 ↗
Evidence and transparency

Sources & review

Official repository ↗Guide documentation ↗
Source record details
Source availability
Active
Sources checked
14 October 2026
Original website title
langfuse/langfuse · Official homepage title · retained in its source language
How we review sourcesSuggest a correction