Token Usage Guide

Learn how tokens work, how to monitor usage, and tips to optimize your consumption.

What Are Tokens?

Tokens are the units used to measure AI usage. Think of them as the "fuel" that powers AI responses:

  • Both your input (prompt) and output (response) consume tokens
  • 1 token ≈ 4 characters or ≈ 0.75 words in English
  • Different AI models have different token costs
  • Images and file uploads also consume tokens

📊 Example Token Counts

  • • "Hello, world!" = ~4 tokens
  • • A typical paragraph = ~100-200 tokens
  • • A full blog post (1,500 words) = ~2,000 tokens
  • • A complex report (5,000 words) = ~6,500 tokens

Token Costs by Model

ModelInput CostOutput CostBest For
GPT-4o-mini$0.15/1M$0.60/1MQuick tasks, simple queries
GPT-4o$2.50/1M$10/1MGeneral purpose, balanced
GPT-4 Turbo$10/1M$30/1MComplex reasoning, analysis
Claude 3.5 Sonnet$3/1M$15/1MCoding, creative writing
Claude 3 Opus$15/1M$75/1MComplex tasks, research

Monitoring Your Usage

Dashboard View

Your dashboard displays real-time token usage:

  • Current Period Usage: Tokens used this billing cycle
  • Remaining Tokens: Tokens left in your monthly allocation
  • Usage Percentage: Visual progress bar
  • Daily Breakdown: Chart showing usage over time
  • Model Distribution: Which models you're using most

Usage Alerts

Set up alerts to avoid unexpected overage charges:

  • Email notification at 80% of monthly allocation
  • Warning at 90% usage
  • Alert when entering overage territory
  • Daily usage summary emails (optional)

Optimizing Token Usage

💡 Choose the Right Model

Use GPT-4o-mini for simple tasks and reserve powerful models for complex work.

Good: GPT-4o-mini

  • • Spell checking
  • • Simple summaries
  • • Basic Q&A

Better: GPT-4o/Claude

  • • Complex analysis
  • • Creative writing
  • • Code generation

✂️ Be Concise

Shorter prompts use fewer tokens. Be specific but concise.

❌ Inefficient (200 tokens)

"I would like you to please write a very detailed and comprehensive blog post about the benefits of artificial intelligence in modern healthcare, including multiple examples..."

✅ Efficient (50 tokens)

"Write a blog post about AI in healthcare with examples"

🎯 Use Presets

ARKA AI presets are optimized for token efficiency. They include only necessary instructions and use the most cost-effective model for each task.

📄 Optimize File Uploads

Large files consume more tokens:

  • • Extract only relevant sections before uploading
  • • Compress images before processing
  • • Use text files instead of PDFs when possible
  • • Split large documents into smaller chunks

Estimating Token Usage

Here are rough estimates for common tasks:

TaskApprox. TokensCost (GPT-4o)
Simple question & answer100-300$0.0003
Social media post200-500$0.0005
Email (professional)300-800$0.0008
Blog post (500 words)1,000-1,500$0.0015
Blog post (1,500 words)2,500-3,500$0.0035
Document summarization (10 pages)3,000-5,000$0.0050
Code generation (complex)2,000-4,000$0.0040
Image analysis (with GPT-4o)500-1,000$0.0010

Batch Processing Tips

When processing multiple items, optimize for efficiency:

  • Combine requests: Process multiple items in one prompt when possible
  • Use templates: Reuse prompt structures to reduce redundant instructions
  • Schedule during off-peak: Use playbooks to automate batch tasks
  • Cache results: Store frequently used outputs to avoid regenerating

Usage FAQ

How is token usage calculated?

Token usage = input tokens (your prompt + context) + output tokens (AI response). Both are counted toward your monthly allocation.

What happens to unused tokens?

Unused tokens do not roll over to the next billing period. They reset on your monthly renewal date.

What if I exceed my allocation?

You'll be charged overage fees based on the model used. Check the pricing page for overage rates. You can set spending limits in your account settings.

How can I reduce costs?

Use GPT-4o-mini for simple tasks, write concise prompts, use presets, and monitor your daily usage to stay within your allocation.

Next Steps