Token usage and cost tracking in Agentic Ai

Introduction

Tracking tokens and costs helps you understand how much you use and spend when working with AI models.

When you want to keep your AI usage within a budget.

When you need to know how many tokens your input and output use.

When you want to optimize your prompts to save money.

When you want to report usage for billing or analysis.

When you want to compare costs between different AI models.

Syntax

Agentic_ai

tokens_used = count_tokens(prompt_text)
cost = tokens_used * cost_per_token

Tokens are pieces of words that AI models use to understand text.

Cost per token depends on the AI model and provider pricing.

Examples

Count tokens in a simple greeting and calculate cost with a token price of 0.0001.

Agentic_ai

prompt = "Hello, how are you?"
tokens_used = count_tokens(prompt)
cost = tokens_used * 0.0001

Calculate tokens and cost for a question prompt with a different token price.

Agentic_ai

input_text = "Explain machine learning in simple words."
tokens = count_tokens(input_text)
cost = tokens * 0.0002

Sample Program

This program counts tokens by splitting the prompt on spaces and calculates the cost using a fixed price per token.

Agentic_ai

def count_tokens(text):
    # Simple token count by splitting on spaces
    return len(text.split())

prompt = "What is AI and how does it work?"
cost_per_token = 0.00015

tokens_used = count_tokens(prompt)
cost = tokens_used * cost_per_token

print(f"Tokens used: {tokens_used}")
print(f"Cost: ${cost:.5f}")

Important Notes

Token counting can be more complex than splitting by spaces because AI models use special tokenizers.

Always check your AI provider's documentation for exact token counting methods and pricing.

Tracking tokens helps avoid unexpected high costs.

Summary

Tokens are small pieces of text used by AI models to process language.

Counting tokens helps estimate how much you will pay for AI usage.

Tracking usage and cost keeps your AI projects affordable and efficient.