Settings → My Usage is the read-only view of where your tokens and dollars went. It is scoped to just your user: every key you've minted, plus chat usage from the app itself.
Picking a window
The selector at the top picks the time window:
- Last 7 days
- Last 30 days (default)
- Last 90 days
- Last year
Everything below (summary cards, time-series chart, heatmap, history table) re-fetches against the new window. The "Today" and "This Month" panels at the top are fixed; the window selector doesn't touch them.
What the top panels show
Four cards stay pinned regardless of the active tab:
- Today: tokens and request count (or cost, if your org is metered). Resets at UTC midnight.
- This Month: running month-to-date total, with a progress bar against your personal budget cap. Bar turns yellow above 70%, red above 90%.
- Requests (window): total requests in whatever window you've picked.
- Active Days: distinct days you made at least one call this month.
If your org runs on dedicated infrastructure, cost columns disappear and a green banner appears at the top of the page reminding you that token consumption is included. The token counters still tick, so you can use them to plan capacity.
If you cross any of the alert thresholds your org admin has configured, a yellow callout sits between the page header and the panels, showing the current percentage and the threshold message.
Three tabs
Overview: six summary tiles (cost, requests, input tokens, output tokens, average latency, models used) plus a time-series chart of daily cost across the window, and a pie chart breaking down spend by model. The chart is interactive: hover for the exact daily figure.
Activity: a 7×24 heatmap, day-of-week down the side, hour-of-day across the top, cells coloured by request count. Useful for spotting an unattended job hammering the API at 3am, or confirming that your batch is in fact running when you scheduled it. Hover a cell for the count.
History: a paginated table of individual usage rows: timestamp, model, request type, tokens in / tokens out, latency, cost, and the key that issued the call. 50 rows per page. This is the audit trail when you want to know which call cost what.
Pricing semantics
The numbers on this page reflect your spend against the models you've used. Prompt tokens, cached prompt tokens, and completion tokens contribute to the total. See Rate limits & budgets for how this works.
Cost columns may be hidden depending on your org's billing setup. The token counters stay regardless, because tokens are the relevant capacity signal.
How this differs from the org admin Chat Usage page
If you're an org admin, you also see a "Chat Usage" view under the Organization group. That page aggregates everyone in the org and slices by user, team, and key. My Usage is strictly yours; no other user's traffic appears on it, regardless of your role.
Where next
- API keys: see per-key totals on the keys table, and revoke anything that's gone rogue
- Account settings: your monthly budget cap and billing posture
- Rate limits & budgets: how
usageis reported on every API response and how to read it - Usage API: the operator-scoped HTTP endpoints that back these dashboards
