Integrations
LiteLLM SDK
Trace LiteLLM SDK calls in Muster — `litellm.callbacks = ["langfuse_otel"]` and you're done.
This page covers the LiteLLM SDK (the Python library) — for the LiteLLM proxy see LiteLLM Proxy.
The LiteLLM SDK has built-in Muster (langfuse) telemetry. Add a callback, set credentials, and every call goes to Muster.
Setup
pip install litellm opentelemetry-api opentelemetry-sdk opentelemetry-exporter-otlpimport os
os.environ["LANGFUSE_PUBLIC_KEY"] = "pk-lf-..."
os.environ["LANGFUSE_SECRET_KEY"] = "sk-lf-..."
os.environ["LANGFUSE_OTEL_HOST"] = "https://app.getmuster.io" # optional, defaults to USEnable the callback
import litellm
litellm.callbacks = ["langfuse_otel"]Make a call
response = litellm.completion(
model="gpt-3.5-turbo",
messages=[{"role": "user", "content": "Hello!"}],
)Enrich traces with metadata
response = litellm.completion(
model="gpt-3.5-turbo",
messages=[{"role": "user", "content": "Hello!"}],
metadata={
"generation_name": "welcome-message",
"trace_id": "trace-123",
"session_id": "sess-42",
"tags": ["prod"],
},
)The integration constructs the OTEL endpoint from LANGFUSE_OTEL_HOST —
EU, US, Japan, HIPAA, or self-hosted instances are all supported.