Tech Blog by T.S.

How to Get Token Count with LangChain

LangChainPythonAI

Published at: 5/13/2025

How to Get Token Count with LangChain

This article explains how to retrieve token counts and manage costs when interacting with LLMs using LangChain.

Basic Implementation

You can get token counts from the metadata after executing a LangChain chain as follows:

from langchain_core.prompts import PromptTemplate
from langchain_openai import ChatOpenAI
 
# Set up prompt template
prompt = PromptTemplate(
    input_variables=["text"],
    template="Analyze the following text: {text}"
)
 
# Configure LLM
llm = ChatOpenAI(
  model="gpt-4o",
  temperature=0,
)
 
# Create chain
chain = prompt | llm
 
# Execute chain and get token counts
results = chain.invoke({"text": "Text to analyze"})
 
# Get token counts
input_tokens = results.usage_metadata['input_tokens']
output_tokens = results.usage_metadata['output_tokens']
 
print(f"Input tokens: {input_tokens}")
print(f"Output tokens: {output_tokens}")

Important Considerations

  1. Token counting methods may vary depending on the model
  2. Cost calculations need to be adjusted according to the pricing structure of the model being used
  3. Proper error handling is crucial when implementing batch processing

Summary

LangChain makes it easy to retrieve token counts and manage costs.