🔢 Token Counting

Count tokens before making API calls. Estimate costs, validate inputs, and optimize your prompts for efficiency.

Token Counter Demo

Sample Text

23 tokens

Hello, world! This is a sample text to demonstrate how tokenization works with our models.

Quick Reference

characters per token

~0.75

words per token

1 page

≈ 500 tokens

1 book

≈ 100K tokens

Count Tokens with the API

Python

from mythicdot import MythicDot

client = MythicDot()

# Count tokens for a message
response = client.messages.count_tokens(
    model="mythic-4",
    messages=[
        {
            "role": "user",
            "content": "What is the meaning of life?"
        }
    ],
    system="You are a helpful assistant."
)

print(f"Input tokens: {response.input_tokens}")

# Estimate cost before calling
cost_per_million = 3.00  # Example pricing
estimated_cost = (response.input_tokens / 1_000_000) * cost_per_million
print(f"Estimated input cost: ${estimated_cost:.6f}")
                

Token Conversion Reference

Content Type

Approximate Size

Tokens

Short email

~200 words

~250 tokens

Blog post

~1,000 words

~1,250 tokens

Long article

~5,000 words

~6,500 tokens

Code file (500 lines)

~15 KB

~4,000 tokens

Book chapter

~10,000 words

~13,000 tokens

💡 Token Optimization Tips

Avoid redundant whitespace — extra spaces and newlines add tokens
Use concise language in system prompts — fewer tokens = lower cost
Batch related content — one larger request often beats many small ones
Consider summarization for context — compress long histories
Use structured outputs (JSON) — often more token-efficient than prose

Optimize Your Token Usage

Count tokens to estimate costs and improve efficiency.

Tokenization Guide → Cost Management →