Repolode
Submit
langfuse

langfuse/langfuse

Open source LLM engineering platform for tracing, evaluating, managing prompts, and debugging LLM applications, self-hostable.

Open-source alternative to:LangSmithHeliconeWeights & Biases Weave

View on GitHubWebsite
Stars
36,072
Forks
4,026
Open issues
1,077
Last commit
9 hours ago

What is it?

Langfuse is an open source platform for building and operating LLM applications. It records traces of model calls and other application logic such as retrieval, embeddings, and agent actions, so teams can inspect complex logs and user sessions when something goes wrong.

Beyond tracing, Langfuse covers prompt management with versioning and caching, evaluations including LLM-as-a-judge, code evaluators, user feedback, and manual labeling, plus datasets for test sets and experiments. An LLM playground lets developers iterate on prompts and model settings directly from a bad trace.

It can be used as Langfuse Cloud or self-hosted with Docker Compose, a VM, Kubernetes via Helm, or Terraform templates for AWS, Azure, and GCP. Python and JS/TS SDKs, an OpenAPI spec, and integrations with OpenAI, LangChain, LlamaIndex, LiteLLM, and others are provided.

Key features

  • 01Traces LLM calls and surrounding app logic such as retrieval, embeddings, and agent actions.
  • 02Manages prompts with version control and client and server caching.
  • 03Supports LLM-as-a-judge, code evaluators, user feedback, and manual labeling.
  • 04Provides datasets for test sets, benchmarks, and structured experiments.
  • 05Includes an LLM playground for iterating on prompts and model configurations.
  • 06Offers an OpenAPI spec, Python and JS/TS SDKs, and drop-in OpenAI instrumentation.

Use cases

  • Debugging production LLM chatbots by inspecting traces and user sessions.
  • Running regression tests of prompts against a dataset before deployment.
  • Letting non-engineers edit and version prompts without redeploying the app.
  • Hosting an observability platform for LLM data on your own infrastructure.

Quick start

  1. 1

    Clone the Langfuse repository

    git clone --depth=1 https://github.com/langfuse/langfuse.git
  2. 2

    Enter the project directory

    cd langfuse
  3. 3

    Start Langfuse with Docker Compose

    docker compose up

Things to know

  • Self-hosting runs a heavier stack that includes ClickHouse, and the README recommends Kubernetes for production.
  • The license is listed as NOASSERTION in the metadata, so automated license detection is unclear even though the README links an MIT badge.

Topics

#llm-observability#llmops#prompt-management#llm-evaluation#tracing#openai#self-hosted#analytics#autogen#evaluation

FAQ

What is langfuse?
Open source LLM engineering platform for tracing, evaluating, managing prompts, and debugging LLM applications, self-hostable. Langfuse is an open source platform for building and operating LLM applications. It records traces of model calls and other application logic such as retrieval, embeddings, and agent actions, so teams can inspect complex logs and user sessions when something goes wrong.
Is langfuse free to use? What license does it have?
langfuse is open source under the Other license. Check the license terms before using it in commercial products.
Is langfuse still maintained?
Actively maintained. The last commit to langfuse was on Oct 20, 2026.
What can I use langfuse for?
Debugging production LLM chatbots by inspecting traces and user sessions. Running regression tests of prompts against a dataset before deployment. Letting non-engineers edit and version prompts without redeploying the app. Hosting an observability platform for LLM data on your own infrastructure.
How do I install langfuse?
Clone the Langfuse repository: git clone --depth=1 https://github.com/langfuse/langfuse.git Enter the project directory: cd langfuse Start Langfuse with Docker Compose: docker compose up
Where can I get langfuse?
The source code and installation instructions are on GitHub: https://github.com/langfuse/langfuse

Summary generated by AI from the project's README. GitHub stats refresh daily.

Related projects