Langfuse

Langfuse is an open-source LLM engineering platform for tracing, evaluating, and debugging LLM applications.

Step 1: Create a Langfuse API key

In Langfuse, go to your project’s Settings > API Keys and create a new key pair. Copy both the Secret Key and Public Key.

Step 2: Enable Broadcast in OpenRouter

Go to Settings > Observability and toggle Enable Broadcast.

Step 3: Configure Langfuse

Click the edit icon next to Langfuse and enter:

• Secret Key: Your Langfuse Secret Key
• Public Key: Your Langfuse Public Key
• Base URL (optional): Default is https://us.cloud.langfuse.com. Change for other regions or self-hosted instances

Step 4: Test and save

Click Test Connection to verify the setup. The configuration only saves if the test passes.

Step 5: Send a test trace

Make an API request through OpenRouter and view the trace in Langfuse.

Custom Metadata

Langfuse supports rich trace hierarchies and metadata. Use the trace field to customize how your traces appear in Langfuse.

Supported Metadata Keys

Example

1
{
2
  "model": "openai/gpt-4o",
3
  "messages": [{ "role": "user", "content": "Summarize this document..." }],
4
  "user": "user_12345",
5
  "session_id": "session_abc",
6
  "trace": {
7
    "trace_id": "workflow_12345",
8
    "trace_name": "Document Processing Pipeline",
9
    "span_name": "Summarization Step",
10
    "generation_name": "Generate Summary",
11
    "environment": "production",
12
    "release": "2.1.0",
13
    "pipeline_version": "2.1.0"
14
  }
15
}

This creates a hierarchical trace structure in Langfuse:

Document Processing Pipeline (trace)
└── Summarization Step (span)
    └── Generate Summary (generation)

Additional Context

• The user field maps to Langfuse’s User ID for user-level analytics
• The session_id field maps to Langfuse’s Session ID for grouping conversations
• Any additional keys in trace are passed as trace metadata and can be used for filtering and analysis in Langfuse

Privacy Mode

When Privacy Mode is enabled for this destination, prompt and completion content is excluded from traces. All other trace data — token usage, costs, timing, model information, and custom metadata — is still sent normally. See Privacy Mode for details.