Skip to main content
Budget alerts are the emergency brake for your credits. Set thresholds, pick channels, get notified before you accidentally run out.

​
Setting an alert

Settings → Budget alerts → New alert.
  • Threshold — as a percent of the monthly plan allowance (50/75/90/100) or an absolute credit amount
  • Channel — which Notification channel gets pinged (email / Slack / webhook)
  • Scope — workspace-wide (all spend) or per-API-key (specific client)

​
Example

Pro plan, 9 000 monthly credits. You set:
  • 50 % (4 500 credits used) → Slack #alerts
  • 75 % (6 750 credits used) → Slack #alerts + email owner
  • 90 % (8 100 credits used) → Slack #alerts + email + auto-pause new inferences

​
Auto-pause at 100 %

Turn on Auto-pause on exhaustion and we stop serving requests the moment your remaining credits hit zero. New inferences return 402 payment_required until you top up or the monthly allowance renews. This protects against runaway scripts.

​
Topup-triggered alerts

You can also subscribe to low topup balance alerts — useful if your team relies on topups for unexpected usage spikes.

​
Per-API-key alerts

Point-protection: alert when a specific API key’s usage crosses a limit, not the whole workspace. Useful for production keys where unusual spend signals a customer churn, a bad query, or abuse.

​
Viewing history

Settings → Budget alerts → each alert row shows the last 10 fires with timestamp and credit value at the time. Useful for audits and incident post-mortems.

​
Who can manage

Owner, Admin and Billing roles can create and edit alerts. Developers and Viewers can see alerts but not modify.