Cost Centerseat-lensed · spend carries a risk lens

Finance · Cost Center

What every AI agent costs — governed, at-risk, and re-routable.

Turn AI spend visibility into a budget decision: are we on track, who drives cost, how much is ungoverned — and where is spend inefficient enough to re-route by runtime selection?

Bottom line

$684.50 total so far (11% estimated); at the ~$115.57/day run-rate this month projects to ~$3.5k vs a $3.0k budget (16% over). 42% of spend ($290.50) is at-risk — $122.10 of it ungoverned and unchargeable. gpt-5-codex runs 6× the cheapest model per token — re-routable spend.

Do nextAttribute $122.10 ungoverned spend →Govern the unbound agents →Set a per-agent ceiling →

Spend you can't charge back or control

governance-tagged · governanceTaggedSpend()

$290.50 of $684.50 total spend (42%) flows through at-risk agents — posture < 80 or unbound by policy.

at-risk 42% · of which $122.10 (18%) is ungoverned / unchargeable

$122.10

ungoverned spend

18% · no policy binding, no chargeback owner

Attribute for chargeback →

$684.50 incl. est.

total spend

the single trusted number · 11% estimated

$3.5k

projected this month

▲ 16% over $3.0k budget

$11.22/session

cost per session

$684.50 ÷ 61 sessions

54%

from cache

cache-read tokens · cost saved

Budget vs. actual — forecast

$3.0k monthly budget

spent so far

$684.50

projected

$3.5k

budget

$3.0k

At the recent ~$115.57/day run-rate, this month projects to $3.5k — 16% over the $3.0k budget.

Spend by agent

$684.50 total

Claude Code

$214.50

support-copilot

$168.40

Cursor

$96.20

Codex CLI

$78.00

data-pipeline-agent

$52.30

marketing-gpt

$44.10

Claude Desktop

$31.00

governed · at-risk · critical/ungoverned

Spend over time

daily · last 7d

$88.20

05-28

$102.50

05-29

$96.10

05-30

$71.00

05-31

$84.40

06-01

$119.70

06-02

$142.60

06-03

Model efficiency — $ per 1k tokens

runtime-selection lever · re-route to cheaper-per-token

gpt-5-codex

$0.032/1k

claude-opus-4-8

$0.032/1k

gpt-4o

$0.032/1k

claude-sonnet-4-6

$0.031/1k

claude-haiku-4-5

$0.006/1k

gpt-5-codex costs $0.032/1k — 6× the cheapest (claude-haiku-4-5 at $0.006/1k). Move latency-tolerant, high-volume work to the cheaper runtime.

By model

5 models

ModelTokensResponsesCost$/1k

claude-opus-4-8 7.7M1.4k$245.50$0.032

gpt-4o 6.7M4.0k$212.50$0.032

claude-sonnet-4-6 3.1M980$96.20$0.031

gpt-5-codex est2.4M540$78.00$0.032

claude-haiku-4-5 9.2M5.4k$52.30$0.006

By agent

7 agents · $290.50 at-risk

AgentGovernanceRiskTokensResp.Cost

Claude Code governedStandard6.9M1.2k$214.50

support-copilot governedElevated5.6M3.1k$168.40

Cursor governedStandard3.1M980$96.20

Codex CLI estungovernedElevated2.4M540$78.00

data-pipeline-agent governedStandard9.2M5.4k$52.30

marketing-gpt ungovernedCritical1.1M860$44.10

Claude Desktop governedStandard800.0k180$31.00

Each agent's spend joins back to its governance posture — so dollars carry the risk lens. $122.10 of this column is ungoverned (highlighted) and has no chargeback owner.

By user

6 identities

UserTokensResponsesCost

mark@qpoint.io 7.7M1.4k$245.50

svc-support@qpoint.io 5.6M3.1k$168.40

jane@qpoint.io 3.1M980$96.20

dev-7f3@duck.com est2.4M540$78.00

svc-data@qpoint.io 9.2M5.4k$52.30

(proxy-only, no identity) 1.1M860$44.10

est rows are estimated from token counts — some providers (OpenAI / Codex) don't report a billed cost, so we derive it from usage. This is usage-based observation, not a billed invoice. For chargeback, attribute spend per cost center in Showback.

Lensed by seat via ?seat. Derived: governanceTaggedSpend (at-risk / ungoverned $) · cost-per-session · $/1k by model — from app/data/c16-fleet.ts. Budget & projection computed client-side.