Insights

Month-end pace, team showback, and the model-mix lever - Acme sample

Renewal desk

Renewal brief

  • On pace to finish $1,410 under the $6,000 budget.
  • Cost per active developer is $179 across 24 people.
  • Idle seats and dual-tool overlap are about $95 this period.
  • $460 is on API keys with no developer attached.
  • Model right-size lever is about $639 if routine frontier traffic moves to a mid-tier model.
  • If usage-priced lines double and Copilot seats stay flat, this period is $7,386, $1,386 over the $6,000 budget.

Today

$4,287

$1,713 under budget

2x usage price

$7,386

$1,386 over budget

3x usage price

$10,485

$4,485 over budget

Seat lines $1,188. Usage-exposed lines $3,099. Copilot is treated as seat-priced and held flat. Cursor, OpenAI, and Anthropic are treated as usage-exposed, which matches how those bills move when request pricing steps up. Cursor enterprise can still include a seat, so the 2x and 3x columns are a planning range.

Month-end forecast

On track - 77% of calendar pace

On pace to finish $1,410 under the $6,000 budget.

Trailing daily burn

$151.43

Last 7 days

Projected month-end

$4,589.86

2 days left in month

Monthly budget

$6,000

Org budget

Runway at this burn

11.3 days

$1,410.14 under if the burn holds

Forecast = month-to-date spend + trailing daily burn times days remaining. Built from official usage APIs. It does not model a price change or a new agent workflow landing next week.

Team showback

Who owns the dollars, compared with headcount. $460 is usage with no developer on the event, usually shared API keys.

TeamSpendSharevs headcountPosture
Backend
5 devs - $360 each
$1,80042%+$433.21Above headcount
Platform
2 devs - $385 each
$77018%+$223.29In line
Frontend
2 devs - $340 each
$68016%+$133.29In line
DevOps
2 devs - $260 each
$52012%-$26.71In line
Unassigned
3 devs - $19 each
$571%-$763.07Below headcount
Unattributed API keys
No person on the event
$46011%--Unallocated

Model right-size

$2,130 on frontier models (70% of model-tagged spend). Moving routine work down a tier is the lever finance asks about at renewal.

Illustrative monthly savings

$639

Illustrative only: 40% of frontier-model spend moved to a mid-tier model at about a quarter of the rate. Quality still has to be checked on real tasks - this is the dollar size of the lever, not a guaranteed cut.

Largest frontier line: claude-sonnet-4

ModelTierCostShare
claude-sonnet-4Frontier$1,24041%
gpt-4oFrontier$89029%
cursor-smallEfficient$42014%
claude-haikuEfficient$31010%
gpt-4o-miniEfficient$1806%
Live Demo | ForgeMeter