Skip to main content
Every incident sureops works on consumes LLM tokens — for diagnosis, recommendations, and verification. Whether you see that spend in dollars depends on how your organization is set up.
This entire page applies to bring-your-own-key (BYOK) organizations — that’s every production organization today. Demo and trial environments run on sureops-supplied LLM capacity and don’t configure dollar budgets — you see either a usage meter (diagnosis credits used this period) or, before that loads, a “Budget managed by sureops” card, instead of USD figures. A parallel managed mode is designed for the same non-dollar-budget behavior but is not yet available for self-serve signup — see BYOK vs Managed LLM. Skip to Managed / demo mode below for how that behavior works, live today for demo/trial and designed-but-unlaunched for managed.
There are two independent cost controls: incident cost caps, which bound a single incident’s spend and can escalate it if it runs away, and a monthly cap, which bounds the organization’s total spend for the billing month.

​
Incident cost caps

Each of the six lifecycle stages (detection, triage, diagnosis, resolution, verification, closure) has its own configurable cap on wall-clock time, LLM call count, and cost — plus one cap for the incident as a whole. All six rows are editable; defaults are just tuned per stage (diagnosis gets the largest budget, while detection and closure default to zero since they typically involve little to no LLM spend). Admins can override any stage’s cap per-environment. Alongside the caps is an enforcement mode, one of:
Enforce mode doesn’t pause new incidents or halt AI org-wide — it only affects the one incident that breached its cap, and only by escalating it to a human rather than letting it continue unattended.
Configure caps and mode in Settings → Budget (organization default) or Settings → Environments → [environment] → Budget (per-environment override, inheriting from the org default until you set one).

​
Monthly cap

The monthly cap is organization-wide (not per-environment) and tracks total LLM spend across all environments and agents for the current UTC calendar month. Configure it in Settings → Budget:
Clearing this field does not mean uncapped. Your organization’s plan tier has its own ceiling on monthly spend, and clearing the field (or requesting a value above that ceiling) clamps you to your tier’s ceiling rather than removing the cap — you can’t request “no cap” any more than you can request above the ceiling. On the Free tier, that ceiling is 0, so clearing the field there has the same effect as explicitly setting it to 0. See Pricing & tiers for your plan’s ceiling.A cap of 0 is a deliberate kill switch, not a no-op. For BYOK organizations, cap 0 immediately puts the monthly gauge into an at-cap/exhausted state; the only thing that actually changes incident behavior is Auto-downgrade at cap — if you have it enabled, new incidents switch to Supervised mode. With auto-downgrade off, a cap of 0 shows as exhausted but nothing about LLM calls is blocked. This is different for managed/demo organizations — see Managed / demo mode, where hitting the cap can hard-block diagnosis.
The cap resets automatically at the start of each UTC calendar month — there’s no manual “reset now” action, since the window is simply computed from the current date each time the gauge is read.

​
What you see

The Settings → Budget page shows:
  • A budget gauge for monthly spend — used vs. cap, with a warning band at your threshold and an at-cap state
  • The budget settings form — monthly cap, warn threshold, auto-downgrade toggle
  • The per-stage/per-incident cap editor described above
Per-incident cost is also visible on the incident itself, in the Incident Hub’s left Stage Navigator panel — total tokens and cost, broken down by agent (commander, diagnosis, resolution, verification). This widget is gated on the incidents:act capability, so Observers don’t see it.

​
Bring your own key (BYOK)

If you supply your own LLM API key, sureops shows cost using the model provider’s list price — it has no visibility into any negotiated rate you may have with your provider, so the dashboard figure may not match your actual invoice exactly. BYOK keys are configured per-environment in Settings → AI Agents → LLM Provider.

​
Managed / demo mode

If your organization runs on sureops-supplied LLM infrastructure — today that means a demo or trial environment; sureops-managed mode for production organizations is designed but not yet available for self-serve signup, see BYOK vs Managed LLM — dollar figures don’t apply, sureops absorbs the LLM cost instead. Instead you see a usage meter tracking your diagnosis (and, where applicable, premortem) allowance for the period — the usage meter and the “Budget managed by sureops” card are shown one at a time, never together (the card is a fallback shown only until the meter has loaded). The meter’s states are: on track, near cap, at cap, using overage (if your plan allows overage), and Blocked — the state that actually stops managed diagnosis once overage headroom (if any) is exhausted. There’s a single related notification, delivered in-app only, when the organization crosses its metering cap; there’s no email or Slack dispatch for it. It’s only sent to members with the Admin or Owner role.

​
Frequently asked questions

Does the budget limit count toward my sureops subscription cost? No. Cost caps track LLM token spend on the model provider’s API (BYOK) or your metered allowance (managed) — not your sureops subscription charge. The two are billed independently. Can I see spend broken down by team member? Not today. Spend is tracked at the incident and agent level, not per person. Is historical cost data retained? Yes — per-incident cost detail is retained as long as your organization is active.
  • Notifications — where the metering-cap notification and incident-timeline breach warnings show up
  • Roles & Capabilities — Admin role required to edit budgets and caps