🔢 tokmeter

Ultra-fast token counter for files and text - supports GPT models, Claude, and more!

tokmeter is the simplest, fastest, and most beautiful CLI tool for counting tokens in your files. Perfect for developers working with AI models, managing token budgets, and optimizing prompts.

✨ Features

• ⚡ Ultra-simple - No subcommands! Just tokmeter file.txt or tokmeter "text"
• 🚀 Ultra-fast - Built on the fastest JavaScript tokenizer
• 🎯 Accurate - Uses OpenAI's official tokenization (via gpt-tokenizer)
• 🌈 Beautiful output - Colorful, clear, and informative displays
• 📁 Batch processing - Count tokens in files, directories, or entire projects
• 🎛️ Flexible - Support for multiple AI models (GPT-4o, Claude, etc.)
• 💰 Cost estimation - See estimated API costs for your tokens
• 🔧 Customizable - Ignore patterns, file extensions, recursive scanning
• 📊 Multiple formats - Human-readable or JSON output
• 🪶 Lightweight - Minimal dependencies, fast installation

🚀 Quick Start

# Install globally
npm install -g tokmeter

# Count tokens in files
tokmeter file1.txt file2.js

# Count tokens in directories
tokmeter ./src --recursive

# Count tokens in text
tokmeter "Hello, world!"

# Count tokens from stdin
echo "Hello, world!" | tokmeter

# See all options
tokmeter --help

📦 Installation

# Global installation (recommended)
npm install -g tokmeter

# Local installation
npm install tokmeter

# Use without installation
npx tokmeter count myfile.txt

🎯 Usage Examples

Count tokens in files

# Count tokens in a single file
tokmeter README.md

# Multiple files
tokmeter src/app.js src/utils.js

# Entire directory (non-recursive)
tokmeter ./src

# Recursive directory scanning
tokmeter ./src --recursive

# Specific file extensions only
tokmeter ./src --recursive --extensions ".js,.ts,.jsx"

Count tokens in text

# Direct text input
tokmeter "Hello, world! How are you today?"

# From stdin
echo "Your text here" | tokmeter

# From file content
cat myfile.txt | tokmeter

Advanced usage

# Use different AI model
tokmeter ./docs --model gpt-4

# JSON output for automation
tokmeter ./src --json > tokens.json

# Summary only (no file details)
tokmeter ./large-project --summary

# Custom ignore patterns
tokmeter ./ --ignore "*.min.js,dist,build"

# Verbose output
tokmeter ./src --verbose

# List supported models
tokmeter --models

🤖 Supported Models

# See all supported models
tokmeter --models

Currently supported:

• GPT models: gpt-4o, gpt-4o-mini, gpt-4, gpt-4-turbo, gpt-3.5-turbo
• Legacy models: text-davinci-003, text-davinci-002
• Claude models: claude-3-opus, claude-3-sonnet, claude-3-haiku

📊 Sample Output

$ tokmeter README.md src/app.js src/utils.js

🔢 TOKMETER - File Token Counter
Using model: gpt-4o
──────────────────────────────────────────────────

📄 File Details:
  /home/user/project/README.md
    Tokens: 1,247 | Size: 4.2 KB
  /home/user/project/src/app.js
    Tokens: 892 | Size: 3.1 KB
  /home/user/project/src/utils.js
    Tokens: 445 | Size: 1.8 KB

📊 Summary:
  Files processed: 3
  Total tokens: 2,584
  Estimated cost: $0.0065

🛠️ CLI Reference

Usage

tokmeter [inputs...] [options]

inputs: Files, directories, or text to count tokens in (reads from stdin if empty)

Options

Option	Description	Default
-m, --model <model>	Model for tokenization	gpt-4o
-r, --recursive	Scan directories recursively	false
-e, --extensions <exts>	File extensions to include	Auto-detect
-i, --ignore <patterns>	Patterns to ignore	node_modules,*.min.js,.git,dist,build
-s, --summary	Show summary only	false
-j, --json	Output as JSON	false
-v, --verbose	Verbose output	false
--models	List supported models	-

🔧 Programmatic Usage

You can also use tokmeter as a library in your Node.js projects:

const { countTokensInText, countTokensInFiles } = require('tokmeter')

// Count tokens in text
const result = countTokensInText('Hello, world!', 'gpt-4o')
console.log(`Tokens: ${result.tokens}`)

// Count tokens in files
const fileResults = await countTokensInFiles(['./src'], {
  model: 'gpt-4o',
  recursive: true,
  extensions: ['.js', '.ts']
})
console.log(`Total tokens: ${fileResults.summary.totalTokens}`)

API Reference

countTokensInText(text, model)

• text string - Text to count tokens in
• model string - Model to use (default: 'gpt-4o')
• Returns object - Result with token count, characters, and cost estimate

countTokensInFiles(paths, options)

• paths string[] - Array of file or directory paths
• options object - Configuration options
  • model string - Model to use
  • recursive boolean - Scan directories recursively
  • extensions string[] - File extensions to include
  • ignore string[] - Patterns to ignore
  • verbose boolean - Enable verbose output
• Returns Promise<object> - Results with file details and summary

💡 Use Cases

• AI Development - Count tokens before sending to APIs
• Cost Management - Estimate API costs for large documents
• Content Analysis - Analyze token distribution in codebases
• Prompt Engineering - Optimize prompts within token limits
• Documentation - Track documentation size and complexity
• Code Review - Understand token impact of changes

🎨 Why tokmeter?

Feature	tokmeter	Others
Speed	⚡ Ultra-fast	🐌 Slow
Output	🌈 Beautiful	📝 Plain text
Models	🤖 10+ models	🤖 Limited
Cost estimation	💰 Built-in	❌ Missing
File handling	📁 Advanced	📄 Basic
CLI experience	✨ Modern	🔧 Basic

🔍 File Type Support

tokmeter automatically detects and processes these file types:

Programming: .js, .ts, .jsx, .tsx, .py, .java, .cpp, .c, .h, .go, .rs, .php, .rb, .swift, .kt, .scala, .cs, .vb

Web: .html, .css, .json, .xml

Data: .yaml, .yml, .sql

Documentation: .md, .txt

Scripts: .sh, .bash

Others: .r, .m

🤝 Contributing

We love contributions! Here's how you can help:

1. 🐛 Report bugs - Found an issue? Open an issue
2. 💡 Suggest features - Have an idea? Start a discussion
3. 🔧 Submit PRs - Fix bugs or add features
4. 📖 Improve docs - Help make our documentation better
5. ⭐ Star us - Show your support!

📄 License

MIT © Sandy Mount

🙏 Credits

• Built with gpt-tokenizer - The fastest JavaScript tokenizer
• Inspired by the need for simple, beautiful token counting tools
• Made with ❤️ for the AI developer community

---

⭐ If tokmeter helped you, please star the repo!