Skip to content
SecondStack SecondStack SecondStack Docs

Cost Management

Cost Management keeps consumption-based AI spend predictable. SecondStack tracks usage down to the individual request and enforces budgets at multiple levels, so overages are caught before they become invoices — the safety net that makes pay-for-what-you-use pricing workable at scale.

Total Platform Spend $798 / $1,200
Remaining: $402
Engineering $642 / $1,000
Chat App $380
API Keys $64
Marketing $156 / $200
78% of budget used
Chat App $32
API Keys $112
d.kowalski
Chat App — Standard Spend: $312 · Budget: $450
Chat App — Agent Mode Spend: $47 · Budget: $180
API Keys
VSCode Spend: $142 · Budget: $200
local-testr-rag-agent Spend: $89 · Budget: $150
Remaining unallocated budget $460
Ranked by total spend this period
a.chen
$268
j.martinez
$251
d.kowalski
$189
r.patel
$176
s.nakamura
$162
l.dubois
$148
m.oconnor
$97
k.berg
$82
n.volkov
$64
p.santos
$51
e.wright
$38

Budgets roll up across the platform, per team, and per user, with a breakdown by key.

Budgets at every level

Section titled “Budgets at every level”

Budgets can be set and enforced for:

  • Per user — each person’s total consumption.
  • Per team — a department or group’s shared allowance, with per-member limits.
  • Per API key — each virtual API key, so a single integration cannot drain the whole budget.

Enforcement happens in the LiteLLM gateway at request time. When a request would exceed any applicable budget, it is blocked rather than billed — the limit is a hard stop, not just a report.

Self-service allocation

Section titled “Self-service allocation”

Users allocate their own budget across the Chat App, the Cowork Agent, and any custom API keys, keeping an explicit unallocated reserve. This lets people manage their own consumption without an admin in the loop, while hard limits keep the total safe.

Threshold alerts

Section titled “Threshold alerts”

Configurable threshold alerts warn before a budget is exhausted. When spend crosses a set percentage of a budget (for example, 70%), the platform notifies the user through an in-app banner, a browser notification, and email. Alerts are de-duplicated within each budget period so a single threshold does not generate repeated noise.

Usage analytics

Section titled “Usage analytics”

ControlTower surfaces usage analytics over the gateway’s spend logs and the Chat App’s request logs:

  • KPI cards for total requests, total spend, latency, and error rate, with period-over-period comparison.
  • Usage over time, broken down by user, team, or model.
  • Top users and model mix.
  • Request-level logs — model, user, token counts, duration, and status per call.

Analytics are scoped by role: administrators see the whole deployment, team supervisors see their teams, and users see their own consumption on the User Dashboard.

Section titled “Related”