Langfuse

Agent observability and evaluation

About Langfuse

Open-source platform for tracing, prompt management and evaluation of LLM apps, self-hosted or as a cloud service.

Langfuse is an open-source engineering platform for LLM applications and agents. Its core is tracing: your app sends a record of each request, including model calls, tool calls and nested steps, and Langfuse shows them as a timeline with latency, cost and token use.

Around tracing it adds prompt management, so prompts can be versioned and changed without a code deploy, along with datasets, scores and evaluations that can be run by code, by a model or by human reviewers. It has integrations for common SDKs and frameworks and accepts OpenTelemetry data.

Teams choose it when they want to keep traces on their own infrastructure, since the same software can be self-hosted, or use the hosted cloud. Plan for the storage that full traces need, and decide which fields should be masked before logging personal or customer data.

Details

  • Open source

    Yes

  • Can run on your own servers

    Yes

  • Developer API

    Yes

Not enough info yet on works with MCP. Check the tool's own site.

Listing written by WithAutobot, checked 2026-11-04. Report a correction or request removal. Back to agent observability and evaluation.

More tools in agent observability and evaluation

Arize Phoenix

by Arize AI

Open-source tool for tracing and evaluating LLM apps, built on OpenTelemetry and runnable on a laptop or server.

  • Open source
  • Can run on your own servers

W&B Weave

by Weights & Biases

Weights & Biases toolkit for tracing LLM and agent calls and running evaluations, tied to its experiment tracking.

  • Open source

Helicone

Open-source proxy and dashboard that logs model requests, cost and latency with a one-line change to your API base URL.

  • Open source
  • Can run on your own servers

Opik

by Comet

Open-source tool from Comet for logging traces, running evaluations and testing prompts for LLM apps and agents.

  • Open source
  • Can run on your own servers

promptfoo

by Promptfoo

Open-source command-line tool for testing prompts and agents against test cases and probing them for security issues.

  • Open source
  • Can run on your own servers

Guardrails AI

Open-source framework for checking model inputs and outputs against validators before an app acts on them.

  • Open source