When you use our credits, you're accessing AI models through our API keys. Credits are consumed based on this simple formula:

Base cost × Model cost = Total credits used

Different operations have different base costs:

• - Chatbot operations have a base cost of 5 credits per operation the agent performs
• - Cell-by-cell operations (SAI commands) have a base cost of 1 credit
• - Deep analysis operations (typically when you need an analysis report on something) have a base cost of 2 credits + 0.1 credits per row analyzed (with bulk discounts after 1000 rows)

The model cost varies depending on which AI model you choose. Lighter models like GPT-4o mini are cheaper (this one is exactly 2 credits), while more powerful models like Claude Sonnet cost more.

Examples:

• - Chatbot operation with GPT-4o mini: 5 × 2 = 10 credits per operation
• - SAI command with GPT-4o mini: 1 × 2 = 2 credits per row
• - Deep analysis of 100 rows with GPT-4o mini: (2 + 10) × 2 = 24 credits
• - Chatbot operation with Claude Sonnet: 5 × 5 = 25 credits per operation

Important notes about credits:

• - Operations like creating pivot tables or marking many rows count as ONE operation, not per row
• - If you have a sheet with 1 million rows and ask the AI to delete all rows that are blank, that still counts as 1 operation
• - Deep analysis operations get bulk discounts for large datasets (over 1000 rows)