Get Started with Langfuse Tracing

This quickstart helps you to integrate your LLM application with Langfuse Tracing. It will log a single LLM call to get you started.

If you are looking for other features, see the overview.

Create new project in Langfuse

Create Langfuse account or self-host
Create a new project
Create new API credentials in the project settings

Log your first LLM call to Langfuse

The @observe() decorator makes it easy to trace any Python LLM application. In this quickstart we also use the Langfuse OpenAI integration to automatically capture all model parameters.

Not using OpenAI? Switch to the “Python Decorator + any LLM” tab.

pip install langfuse openai

.env

LANGFUSE_SECRET_KEY="sk-lf-..."
LANGFUSE_PUBLIC_KEY="pk-lf-..."
# 🇪🇺 EU region
LANGFUSE_HOST="https://cloud.langfuse.com"
# �� US region
# LANGFUSE_HOST="https://us.cloud.langfuse.com"

main.py

from langfuse.decorators import observe
from langfuse.openai import openai # OpenAI integration
 
@observe()
def story():
    return openai.chat.completions.create(
        model="gpt-3.5-turbo",
        max_tokens=100,
        messages=[
          {"role": "system", "content": "You are a great storyteller."},
          {"role": "user", "content": "Once upon a time in a galaxy far, far away..."}
        ],
    ).choices[0].message.content
 
@observe()
def main():
    return story()
 
main()

The @observe() decorator makes it easy to trace any Python LLM application. If you do not use any of our native integrations, you can trace any LLM call via @observe(as_type="generation"). Below find an example using the Anthropic SDK.

pip install langfuse anthropic

.env

LANGFUSE_SECRET_KEY="sk-lf-..."
LANGFUSE_PUBLIC_KEY="pk-lf-..."
# 🇪🇺 EU region
LANGFUSE_HOST="https://cloud.langfuse.com"
# �� US region
# LANGFUSE_HOST="https://us.cloud.langfuse.com"

main.py

from langfuse.decorators import observe, langfuse_context
import anthropic
 
anthopic_client = anthropic.Anthropic()
 
# Wrap LLM function with decorator
@observe(as_type="generation")
def anthropic_completion(**kwargs):
  # optional, extract some fields from kwargs
  kwargs_clone = kwargs.copy()
  input = kwargs_clone.pop('messages', None)
  model = kwargs_clone.pop('model', None)
  langfuse_context.update_current_observation(
      input=input,
      model=model,
      metadata=kwargs_clone
  )
 
  response = anthopic_client.messages.create(**kwargs)
 
  # See docs for more details on token counts and usd cost in Langfuse
  # https://langfuse.com/docs/model-usage-and-cost
  langfuse_context.update_current_observation(
      usage_details={
          "input": response.usage.input_tokens,
          "output": response.usage.output_tokens
      }
  )
 
  # return result
  return response.content[0].text
 
@observe()
def main():
  return anthropic_completion(
      model="claude-3-opus-20240229",
      max_tokens=1024,
      messages=[
          {"role": "user", "content": "Hello, Claude"}
      ]
  )
 
main()

npm i langfuse
# or
yarn add langfuse
 
# Node.js < 18
npm i langfuse-node
 
# Deno
import { Langfuse } from "https://esm.sh/langfuse"

.env

LANGFUSE_SECRET_KEY="sk-lf-..."
LANGFUSE_PUBLIC_KEY="pk-lf-..."
# 🇪🇺 EU region
LANGFUSE_BASEURL="https://cloud.langfuse.com"
# 🇺🇸 US region
# LANGFUSE_BASEURL="https://us.cloud.langfuse.com"

Example usage, most of the parameters are optional and depend on the use case. For more information, see the JS/TS SDK docs or end-to-end example notebook.

server.ts

import { Langfuse } from "langfuse";
 
const langfuse = new Langfuse();
 
const trace = langfuse.trace({
  name: "my-AI-application-endpoint",
});
 
// Example generation creation
const generation = trace.generation({
  name: "chat-completion",
  model: "gpt-3.5-turbo",
  modelParameters: {
    temperature: 0.9,
    maxTokens: 2000,
  },
  input: messages,
});
 
// Application code
const chatCompletion = await llm.respond(prompt);
 
// End generation - sets endTime
generation.end({
  output: chatCompletion,
});

The integration is a drop-in replacement for the OpenAI Python SDK. By changing the import, Langfuse will capture all LLM calls and send them to Langfuse asynchronously (for Azure OpenAI, use from langfuse.openai import AzureOpenAI).

pip install langfuse

.env

LANGFUSE_SECRET_KEY="sk-lf-..."
LANGFUSE_PUBLIC_KEY="pk-lf-..."
# 🇪🇺 EU region
LANGFUSE_HOST="https://cloud.langfuse.com"
# �� US region
# LANGFUSE_HOST="https://us.cloud.langfuse.com"

.py

- import openai
+ from langfuse.openai import openai
 
Alternative imports:
+ from langfuse.openai import OpenAI, AsyncOpenAI, AzureOpenAI, AsyncAzureOpenAI

Use the OpenAI SDK as you would normally. Example:

completion = openai.chat.completions.create(
  model="gpt-3.5-turbo",
  messages=[
      {"role": "system", "content": "You are a very accurate calculator."},
      {"role": "user", "content": "1 + 1 = "}],
)

.env

LANGFUSE_SECRET_KEY="sk-lf-..."
LANGFUSE_PUBLIC_KEY="pk-lf-..."
# 🇪🇺 EU region
LANGFUSE_BASEURL="https://cloud.langfuse.com"
# 🇺🇸 US region
# LANGFUSE_BASEURL="https://us.cloud.langfuse.com"

With your environment configured, call OpenAI SDK methods as usual from the wrapped client.

import OpenAI from "openai";
import { observeOpenAI } from "langfuse";
 
const openai = observeOpenAI(new OpenAI());
 
const res = await openai.chat.completions.create({
  messages: [{ role: "system", content: "Tell me a story about a dog." }],
  model: "gpt-3.5-turbo",
  max_tokens: 300,
});

The integration uses the Langchain callback system to automatically capture detailed traces of your Langchain executions.

pip install langfuse

# Initialize Langfuse handler
from langfuse.callback import CallbackHandler
langfuse_handler = CallbackHandler(
    secret_key="sk-lf-...",
    public_key="pk-lf-...",
    host="https://cloud.langfuse.com", # 🇪🇺 EU region
  # host="https://us.cloud.langfuse.com", # 🇺🇸 US region
)
 
# Your Langchain code
 
# Add Langfuse handler as callback (classic and LCEL)
chain.invoke({"input": "<user_input>"}, config={"callbacks": [langfuse_handler]})

Also works for run and predict methods.

chain.run(input="<user_input>", callbacks=[langfuse_handler]) # Legacy
conversation.predict(input="<user_input>", callbacks=[langfuse_handler])

For more details, see the Langchain integration docs.

The integration uses the Langchain callback system to automatically capture detailed traces of your Langchain executions.

npm i langfuse-langchain

import { CallbackHandler } from "langfuse-langchain";
// Deno: import CallbackHandler from "https://esm.sh/langfuse-langchain";
 
const langfuseHandler = new CallbackHandler({
  secretKey: "sk-lf-...",
  publicKey: "pk-lf-...",
  baseUrl: "https://cloud.langfuse.com", // 🇪🇺 EU region
  // baseUrl: "https://us.cloud.langfuse.com", // 🇺🇸 US region
});
 
// Your Langchain code
 
// Add Langfuse handler as callback to `run` or `invoke`
await chain.invoke({ input: "<user_input>" }, { callbacks: [langfuseHandler] });

For more details, see the Langchain integration docs.

The integration uses the LlamaIndex callback system to automatically capture detailed traces of your LlamaIndex executions.

pip install llama-index langfuse

from llama_index.core import Settings
from llama_index.core.callbacks import CallbackManager
from langfuse.llama_index import LlamaIndexCallbackHandler
 
langfuse_callback_handler = LlamaIndexCallbackHandler(
    public_key="pk-lf-...",
    secret_key="sk-lf-...",
    host="https://cloud.langfuse.com"
)
Settings.callback_manager = CallbackManager([langfuse_callback_handler])

Traces and metrics from your LlamaIndex application are now automatically tracked in Langfuse. If you construct a new index or query an LLM with your documents in context, your traces and metrics are immediately visible in the Langfuse UI.

The low-level SDK gives you full control over the traces logged to Langfuse. For a less verbose integration, consider using the @observe() decorator.

pip install langfuse

.env

LANGFUSE_SECRET_KEY="sk-lf-..."
LANGFUSE_PUBLIC_KEY="pk-lf-..."
# 🇪🇺 EU region
LANGFUSE_HOST="https://cloud.langfuse.com"
# �� US region
# LANGFUSE_HOST="https://us.cloud.langfuse.com"

Example usage, most of the parameters are optional and depend on the use case. For more information, see the python docs.

server.py

from langfuse import Langfuse
 
# Create Langfuse client
langfuse = Langfuse()
 
# Create generation in Langfuse
generation = langfuse.generation(
    name="summary-generation",
    model="gpt-3.5-turbo",
    model_parameters={"maxTokens": "1000", "temperature": "0.9"},
    input=[{"role": "system", "content": "You are a helpful assistant."}, {"role": "user", "content": "Please generate a summary of the following documents \nThe engineering department defined the following OKR goals...\nThe marketing department defined the following OKR goals..."}],
    metadata={"interface": "whatsapp"}
)
 
# Execute model, mocked here
# chat_completion = openai.ChatCompletion.create(model="gpt-3.5-turbo", messages=[{"role": "user", "content": "Hello world"}])
chat_completion = "completion":"The Q3 OKRs contain goals for multiple teams..."
 
# Update span and sets end_time
generation.end(output=chat_completion)
 
# The SDK executes network requests in the background.
# To ensure that all requests are sent before the process exits, call flush()
# Not necessary in long-running production code
langfuse.flush()