count-tokens

count-tokens

A simple CLI tool to count tokens in files or clipboard content using OpenAI's tiktoken library.

🔗 GitHub Repository

Installation

npm install -g count-tokens

or with bun:

bun add -g count-tokens

Usage

count-tokens [file] [options]

Options

• -m, --model <model> - OpenAI model to use for encoding (default: "gpt-4")
• -e, --encoding <encoding> - Specific encoding to use (overrides model)
• -d, --details - Show detailed token information including character count and cost estimates
• -c, --chunks <size> - Split output into chunks of specified token size
• --clipboard - Count tokens from clipboard content instead of a file
• -h, --help - Display help
• -V, --version - Display version

Examples

Count tokens in a file using the default GPT-4 encoding:

count-tokens myfile.txt

Count tokens using GPT-3.5 Turbo encoding:

count-tokens myfile.txt --model gpt-3.5-turbo

Show detailed information including cost estimates:

count-tokens myfile.txt --details

Show chunk breakdown for 4096 token chunks:

count-tokens myfile.txt --chunks 4096

Use a specific encoding directly:

count-tokens myfile.txt --encoding cl100k_base

Clipboard Examples

Count tokens from clipboard content:

count-tokens --clipboard

Count clipboard tokens with GPT-3.5 Turbo encoding:

count-tokens --clipboard --model gpt-3.5-turbo

Show detailed information for clipboard content:

count-tokens --clipboard --details

Copy text and immediately count tokens (macOS):

echo "Your text here" | pbcopy && count-tokens --clipboard

Copy text and immediately count tokens (Linux):

echo "Your text here" | xclip -selection clipboard && count-tokens --clipboard

Supported Models

• gpt-4, gpt-4-32k
• gpt-3.5-turbo
• gpt-4o, gpt-4o-mini
• text-davinci-003
• text-embedding-ada-002
• And many more OpenAI models

Supported Encodings

• gpt2
• cl100k_base
• o200k_base
• p50k_base
• p50k_edit
• r50k_base

Development

bun install
bun run index.ts <file>

License

MIT