Skip to content

Usage & throttle

Usage & throttle is a fully read-only usage-and-consumption dashboard: an Active throttles panel (what’s being rate-limited right now), usage-vs-limit quota (requests/tokens/cost against the authored per-subject quotas), a requests/throttled trend, requests by provider, token consumption by model, and a regroupable usage-breakdown table (by model or provider). Per-account usage-alert thresholds are authored on the Alerts surface in System, not here.

The Usage & throttle panel showing the Active throttles panel at the top, summary tiles for requests/tokens/cost/throttled, a usage-vs-limit quota card, and requests/tokens charts.
Monitoring & Reporting → Usage & throttleCitadel synthetic data on a non-production Arbitex demo instance. Never client data.
  1. Open Monitoring & Reporting → Usage & throttle. Active throttles leads the page — a calm “no active throttles” state, an honest “temporarily unavailable” note if the platform enforcement-plane read can’t be reached, or an urgent panel naming every throttled subject, its trigger (for example “@110%”), the action taken, effective-until, and the breached limit.

  2. Review the summary tiles (Requests, Tokens, Estimated cost, Throttled) and the Usage quota card (reset date, billing period, and used/limit meters for requests, tokens, and cost — a missing limit reads “(no limit),” never a fabricated denominator).

  3. Pick a Trend period for the requests-over-time chart.

  4. Review Requests by provider and Token consumption by model.

  5. Use the usage-breakdown table’s Group by control to regroup by model or provider (requests, tokens, and estimated cost per group).

  6. Scroll to the retained Recent usage rollups table for the raw per-window figures.

Terminal window
curl -X POST "https://admin.arbitex.ai/api/v1/admin/monitoring/usage/quota" \
-H "Authorization: Bearer $ADMIN_TOKEN" \
-H "Content-Type: application/json" \
-d '{}'

Trend and breakdown charts read usage/timeseries and usage/breakdown; active enforcement reads usage/active-throttles; the raw rollup table reads usage/list.

Available on every plan. Quota limits, if any, are configured per subject elsewhere in the admin console — this page only shows usage against them.

Active-throttle state is a platform enforcement-plane read; in air-gapped deployments without a path to the platform, it shows “temporarily unavailable” rather than an assumed-clear state.