AI Usage
How many tokens the company used, who used them, and with which agent and model. Use it to estimate the provider’s bill.
Overview
Where to find itSettingsAI SettingsAI Usage
- Choose the period at the top: today, 7 days, 30 days, all time, or custom.
- Total tokens = input tokens (the question and the data the assistant read) + output tokens (the answer). Cache Read are tokens reused from an earlier request, which cost less.
- Filter by source (chat or scheduled run), user, agent, provider and model.
- The table lists each activity with its user, agent, model, requests, tokens and last time.
Notes
- Most tokens go to input, not the answer: every step sends the tool list and the data read so far. One question about top products used over 100K tokens.