Team Console — Usage & Cost

Cost shows where a team’s spend concentrates, its model mix and efficiency, inefficiency signals, and a per-user cost ranking. Use it to see where the money goes and where you can trim it; it works for a single team or across all managed teams.

Principle banner

A cost ≠ performance banner and a cost-estimate disclaimer sit at the top. High cost doesn’t mean someone did poor work — harder, more complex tasks simply cost more — so treat the figures here as a way to see where spend goes, not as a way to rank people or teams.

KPI row

The top-line numbers that sum up team spend: Spend (with a delta versus the previous equal-length window), Tokens (total usage), and the average per user (spend divided by users). If the delta jumps, look to the sections below for why.

Where cost concentrates

Points out where spend piles up: the user who sent the most requests, the top-cost user, and the team’s most-used model and its share — so you can tell at a glance whether cost is skewed toward one person or one model.

Usage & cost trend

Shows the trend over the period. Toggle between cost and tokens to see whether spend is rising or falling and whether any day spikes.

Model section

How much you use each model, and how efficiently. A mix bar and share list show the model makeup, a model-usage trend shows change over time, and per-model token efficiency gives tokens and $ per API request (a turn may make several requests) — higher values mean you’re leaning on that model more heavily.

Inefficiency

Flags patterns that can turn into waste, by count rather than dollars: the rework rate, correction turns broken down by reason (negation / revert / reissue / error-paste), and cost-spike user cards for anyone whose spend jumped. Each is a lead on why cost went up.

Cost by user

A bar ranking of users by cost. Click a bar for a drill modal — grade, completion, requests/tokens, previous cost, and a work summary — so you can see what the high-cost users actually worked on.

Cross-team comparison

Appears in the all-teams view when you manage two or more teams, comparing them side by side by total or trend. Use it when you want to see which team spend concentrates in.

Cost is an estimate and can be partial

Costs are estimated at public list price, and the data can be partial. On a partial read, Spend shows — with a note and the delta is suppressed; when pricing coverage is merely incomplete, Spend shows the priced-only figure with a coverage note. Token and model shares are withheld (shown as —) rather than treated as zero unless every value is present, and the per-user ranking is only sorted and shown when cost data is complete enough to rank. When some teams fail to load, a note reports how many are missing.