📊 CSV Upload Guide

Learn how to prepare your data for analysis

📋 Required Columns

model
Text (required)
The AI model name (e.g., "gpt-4", "claude-3-opus", "gemini-pro")
prompt_tokens
Number (required)
Number of input tokens used
completion_tokens
Number (required)
Number of output tokens generated
team
Text (optional)
Team or department name (if not provided, we auto-generate from API key, email, or customer ID)

Optional Columns for Auto-Team Mapping

If you don't have a team column, we can auto-generate it from:

api_key
Maps unique API keys to teams (Team-A, Team-B, etc.)
user_email
Extracts team from email domain ([email protected] → Marketing)
customer_id
Creates team names from customer IDs (Customer-123, Customer-456)

🤖 Supported AI Models

OpenAI

Model Name Type Input Price Output Price
gpt-4 Advanced $0.03 / 1K tokens $0.06 / 1K tokens
gpt-4-turbo Fast Advanced $0.01 / 1K tokens $0.03 / 1K tokens
gpt-3.5-turbo Budget $0.0005 / 1K tokens $0.0015 / 1K tokens

Anthropic (Claude)

Model Name Type Input Price Output Price
claude-3-opus Advanced $0.015 / 1K tokens $0.075 / 1K tokens
claude-3-sonnet Balanced $0.003 / 1K tokens $0.015 / 1K tokens
claude-3-haiku Budget $0.00025 / 1K tokens $0.00125 / 1K tokens

Google (Gemini)

Model Name Type Input Price Output Price
gemini-1.5-pro Advanced $0.00125 / 1K tokens $0.00375 / 1K tokens
gemini-1.5-flash Budget $0.000075 / 1K tokens $0.0003 / 1K tokens
gemini-pro Standard $0.00025 / 1K tokens $0.0005 / 1K tokens

Azure OpenAI

Model Name Equivalent Input Price Output Price
azure-gpt-4 GPT-4 $0.03 / 1K tokens $0.06 / 1K tokens
azure-gpt-4-turbo GPT-4 Turbo $0.01 / 1K tokens $0.03 / 1K tokens
azure-gpt-35-turbo GPT-3.5 Turbo $0.0005 / 1K tokens $0.0015 / 1K tokens

📝 Example CSV Format

Minimal Format (with auto-team mapping)

model,prompt_tokens,completion_tokens,api_key gpt-4,150,250,sk_prod_123 gpt-3.5-turbo,50,100,sk_prod_456 claude-3-opus,200,300,sk_prod_123

With Team Column

model,prompt_tokens,completion_tokens,team gpt-4,150,250,marketing gpt-3.5-turbo,50,100,engineering claude-3-opus,200,300,data_science

From Email Domain

model,prompt_tokens,completion_tokens,user_email gpt-4,150,250,[email protected] gpt-3.5-turbo,50,100,[email protected] gemini-pro,75,150,[email protected]

💡 Tips for Best Results

✓ Data Format Tips

  • Token counts must be whole numbers (no decimals)
  • Model names are case-insensitive (GPT-4, gpt-4, Gpt-4 all work)
  • Model name variations are supported (gpt-4-0613 → gpt-4)
  • CSV encoding should be UTF-8 for special characters
  • Maximum 100,000 rows per upload

📊 Data Quality Tips

  • Include at least 1 month of data for meaningful trends
  • Include multiple teams if comparing cost allocation
  • Include variety of models to see optimization opportunities
  • Consistent team naming helps identify patterns

⚠️ Common Issues

❌ Error: Missing required columns

  • Check that you have: model, prompt_tokens, completion_tokens
  • Column names are case-sensitive in error messages but flexible in matching
  • Try: prompt_token, input_tokens instead of prompt_tokens

❌ Error: File too large

  • Maximum file size is 10MB
  • Split large datasets into multiple uploads
  • Remove unnecessary columns

❌ Error: Invalid token count

  • Token counts must be whole numbers (0, 150, 250, etc.)
  • Remove decimals (150.5 → 150)
  • Remove text like "tokens" or "k"

✅ Success: What happens next?

After upload, you'll see:

  • ✓ Total cost breakdown by team
  • ✓ Cost by provider (OpenAI, Claude, Gemini, etc.)
  • ✓ Waste patterns detected
  • ✓ Cost-saving recommendations
  • ✓ Team attribution accuracy

📥 Get Started

Download a template CSV file to get started. You can edit it with your data and upload it.

📤 Go to Uploader ← Back to Home