Review Usage and Cost
The Usage page in the administration area reports token cost across every user, application, and workspace. Open the Admin section and select Usage.
Read this page to see where cost concentrates before you decide a resource request or guide a user toward a lighter model. Filters narrow the view, the Tokens section breaks the cost down, and Top Usage ranks the users with the highest cost. Opening the Admin section needs administrator access.
Filter across users
Five controls narrow every section on the page:
- Time range: 7 days, 30 days, 90 days, 1 year, or a range you set.
- All users: narrows to the users you pick.
- All applications: narrows to the applications you pick.
- All workspaces: narrows to the workspaces you pick.
- All models: narrows to the models you pick.
The All users filter widens the view to the whole organization. Pick specific people, or leave it set to everyone.
Token cost and usage
The Tokens card carries a Cost and a Usage view. Cost shows the estimated dollar cost across the users you selected, and Usage shows the token counts. Both plot the range as a chart, with one color for each model.
Usage reports four figures: Total Tokens, Input Tokens, Output Tokens, and Cached Tokens. Cached tokens are reasoning together with cache reads and writes. Each user's LLM Budget is a dollar amount, so compare budgets with the Cost view, not with token counts.
Check where usage concentrates
The Top Usage section ranks the heaviest consumers in three tables: User, Workspace, and Application. Each table lists the top ten by Est. Cost or token count. A single By Cost and By Tokens toggle sets the measure for all three, and a time range sets the period.
Usage concentrates heavily, so a few users, workspaces, and applications usually account for most of the cost. Read these rankings before judging whether one user's request is reasonable.