Skip to main content

Token Management

Understanding AI token economics and optimization.

What Are Tokens?​

Tokens are atomic units of text that AI models process.

"Hello world" ≈ 2-3 tokens
Typical word ≈ 1.3 tokens

Token Pricing​

Different models charge different rates:

Model           Input   Output  Best For
─────────────────────────────────────────
Claude Haiku    $0.80   $4      Budget
Gemini Pro      $0.50   $1.50   Speed
Claude Sonnet   $3      $15     Balanced
Claude Opus     $15     $75     Power
GPT-4 Turbo     $10     $30     Advanced

Optimization Strategies​

1. Right-Size the Model​

Simple task: Use Gemini ($0.50/M)
Complex task: Use Opus ($15/M)

2. Optimize Prompts​

Verbose prompt:  12,000 tokens  Cost: $0.03
Concise prompt:  2,000 tokens   Cost: $0.005
Savings: 83%!

3. Cache Results​

First run with 30k tokens
Second time with cached result: 2k tokens
Savings across team: massive

Full Details: See Token Tracking Guide