token-analyzer 0.1.2

A Python library for analyzing and calculating token usage costs for various LLM models like OpenAI and Anthropic, with support for real-time currency conversion.

Token Analyzer Library

TokenAnalyzer is a Python library to analyze and calculate token usage costs for various LLM models such as OpenAI, Anthropic, and others. The library supports live currency conversion using the FreeCurrencyAPI or uses default exchange rates if the API key is not provided.

Features

• Calculate costs for input/output tokens across various LLM models.
• Supports multiple currencies (INR, USD, EUR, etc.).
• Can write token usage information to a CSV file.
• Logging options for detailed analysis of costs.
• Use FreeCurrencyAPI for real-time currency conversion.

Installation

You can install the library directly from PyPI (after publishing):

pip install token-analyzer

Usage

Here is a basic usage example:

from analyzer.token_analysis import TokenAnalysis

# Response data from a model
response = {
    'usage': {'prompt_tokens': 1000, 'completion_tokens': 500},
    'model': 'claude-2.1'
}

# Initialize the analyzer
analyzer = TokenAnalysis(output_currency="INR", output_csv="output/token_costs.csv", create_csv=True, log_info=True)

# Calculate token cost
total_cost = analyzer.token_analysis(response, "ANTHROPIC")
print(f"Total Cost: {total_cost}")

Example without CSV output and logging disabled:

analyzer = TokenAnalysis(output_currency="USD", create_csv=False, log_info=False)
total_cost = analyzer.token_analysis(response, "OPENAI")
print(f"Total Cost: {total_cost}")

License

This project is licensed under the MIT License - see the LICENSE file for details.

Links

• GitHub Repository