What this page is for
See how much each AI feature is used, what it is likely to cost, and which calls failed and why.
Who can use it
See AI usage and costs (by default Admin and Finance).
How to read the month
- The page opens on the current month in your timezone. Use Previous month and Next month to move.
- The first table totals calls, failed calls, tokens and estimated cost by feature and connection.
- Latest calls shows the last 50 calls with the model, how long they took and any error.
Fields and columns
| Column | Meaning |
|---|---|
| Feature | What asked the AI. "Connection test" is the Test button. |
| Connection | Which connection answered or failed. |
| Calls / Failed | Number of attempts, and how many failed. A failed call was retried on the next connection. |
| Tokens in / out | Text sent and received, in tokens (roughly four letters each). |
| Estimated cost | From the prices on the connection. Local AI is free. |
| Time | How long the answer took. |
| Result | OK, or the error from the service. |
Statuses and automation
Every AI call is recorded here automatically. Monthly budgets on connections use these estimates.
Settings that change this page
Prices and budgets are set on each connection under AI connections.
FAQs
Why is the cost different from my bill? It is an estimate from the prices you entered. The bill from the service is final.
Is the text of conversations stored here? No. Only counts, times and errors.
Troubleshooting
| Problem | What to do |
|---|---|
| Costs show as 0 | Enter the prices per million tokens on the connection. |
| Many failed calls | Open the error, then test the connection under AI connections. |