Usage

The Usage page shows who is consuming AI resources and how much it costs. You can view data aggregated by team or drilled down to individual users. Both views include the same core metrics — requests, tokens, cost — but at different levels of granularity.

This page is designed for team leads who need to understand their team's AI consumption, and for engineering managers who want to compare usage patterns across teams or identify individual cost outliers.

Views

Two toggle buttons at the top of the page switch between:

By user — shows individual users, each with their team affiliation
By team — shows aggregated metrics per team

The rest of the page — KPI cards, charts, and breakdown table — updates to reflect the selected view.

Filters

Both views share a set of filters below the view toggle:

Filter	What it does
Time range	Select a window: 24h, 7d, 14d, 30d, or Custom.
API keys	Filter to requests made with specific API keys. Defaults to all keys.
Tags	Filter by tags attached to requests.
Teams	Filter to specific teams (available in both views — in the user view, this filters to users belonging to the selected teams).

Live Mode

Toggle Live Mode on to auto-refresh the page as new requests arrive. The data updates continuously without a manual page reload. Turn it off when you need a stable snapshot for reporting or screenshots.

KPI cards

Four summary cards appear at the top:

Card	What it shows
Active Users	Distinct users who made at least one request in the selected range
Avg Requests / User	Total requests divided by active users
Requests	Total API requests
Total Tokens	Sum of input and output tokens across all requests
User Cost	Total cost attributed to the filtered user set

Activity chart

A bar chart visualizes activity over the selected time range:

By user: shows the count of active users per day, giving you a sense of daily engagement.
By team: shows a stacked bar chart colored by team, so you can see which teams drive the most traffic on any given day.

Breakdown

The table at the bottom of the page provides the detail. Each row represents a user or team, depending on the selected view.

Column	Description
User Name	Name and email. A colored badge shows the user's team.
Trend	Sparkline showing request volume over the selected time range.
Models	List of models the user has called, with a "+N" overflow indicator if more than three.
Requests	Total request count. Sortable.
Input	Total input tokens. Sortable.
Output	Total output tokens. Sortable.
Cost	Total cost. Sortable.

Click any column header to sort. Use the search box above the table to filter by name or email (user view) or team name (team view).

Practical uses

Track per-team spend: switch to the team view, set a 30d range, and sort by Cost to see which teams drive the most spend. Share this with finance or leadership to allocate AI budgets.
Identify heavy consumers: in the user view, sort by Requests or Cost to find individuals with outsized usage. This isn't about policing — it often reveals power users whose workflows could inform best practices for the rest of the team.
Compare model preferences: the Models column shows which models each team or user gravitates toward. If one team relies heavily on an expensive model, consider whether a cheaper alternative would serve them just as well.
Catch anomalies: a sudden spike in a user's trend sparkline may indicate a runaway script or misconfigured agent. Live Mode makes it easy to spot these in real time.