Skip to content
Pydantic Docs

LangChain

See every step your LangChain chains and LangGraph agents take: the model calls, tool calls, how many tokens (the units a model reads and bills by, a few characters of text each) they used, how long they took, and any errors, as a trace (the full journey of one request, made of nested spans, where each span is one unit of work with a name, a start, and a duration) in Logfire.

LangChain has built-in tracing through LangSmith that speaks OpenTelemetry (the open standard Logfire receives data in), so there’s no separate Logfire instrument call. You turn on that tracing with a few environment variables, and Logfire receives the data.

What you’ll capture

  • Each chain, agent, and tool step as a span, with its duration and any exceptions
  • Model calls, with the conversation and the number of tokens used
  • Nested steps of a chain or agent, shown as child spans in one trace

Before you start

You’ll need a Logfire project. Open Add data in your project (top navigation) and follow the setup for your language: it signs your machine in with logfire auth (a browser sign-in, no token to copy) and, for production or other languages, creates a write token (the credential your app uses to send data). New to Logfire? Start with Getting Started.

Installation

Install logfire:

Terminal
pip install logfire

This works with your existing langchain (or langgraph) install: nothing extra to add.

Usage

Set three environment variables to turn on LangSmith’s OpenTelemetry tracing, call logfire.configure() to connect to your project, and Logfire receives the data. The variables must be set before you import langchain or langgraph.

LANGSMITH_OTEL_ENABLED=true
LANGSMITH_OTEL_ONLY=true
LANGSMITH_TRACING=true

The example below calls an OpenAI model, so it also needs your model provider’s credential: set the OPENAI_API_KEY environment variable to your OpenAI API key (swap in another provider’s model and key if you prefer).

Here’s a complete example using LangGraph:

import os

import logfire

# These environment variables need to be set before importing langchain or langgraph
os.environ['LANGSMITH_OTEL_ENABLED'] = 'true'
os.environ['LANGSMITH_OTEL_ONLY'] = 'true'
os.environ['LANGSMITH_TRACING'] = 'true'

from langchain.agents import create_agent

logfire.configure()


def add(a: float, b: float) -> float:
    """Add two numbers."""
    return a + b


math_agent = create_agent('openai:gpt-5-mini', tools=[add], name='math_agent')

result = math_agent.invoke({'messages': [{'role': 'user', 'content': "what's 123 + 456?"}]})
print(result['messages'][-1].content)

The resulting trace looks like this in Logfire:

Logfire LangChain Trace

Verify it worked

Run your program, then open your project in the Logfire web app and go to the Live view. Within a few seconds you should see a trace for the agent run, with a span for each step. Click into it to see the model calls and tool calls.

Troubleshooting

Not seeing your LangChain steps in Logfire? Check these first:

  • The three LANGSMITH_* environment variables are set before importing langchain or langgraph. Set them at the very top of your program, above the imports.
  • logfire.configure() runs at startup, so the connection is ready when the first step runs.
  • Your Logfire write token is set. In local development, run logfire projects use <your-project>; in production, set the LOGFIRE_TOKEN environment variable. See Getting Started.

Reference