Token usage — CodeAgent Mobile Wiki
The CodeAgent Mobile Wiki

Every feature, explained — organized by the path you actually take: set up, launch a session, supervise the run, ship the PR.

Free

Token usage

See exactly what cost-saving saved on a session — tokens, dollars, and which model it came from.

01 · RUNActive
Agent works
  • Cost-saving is on
02 · MEASUREMeasured
Savings are measured
  • On your own machine or box
03 · OPENReport
⋯ → Token usage
  • From Home, Sessions, or in-session
04 · READExported
Totals, chart, models
  • Export the window as CSV

Once cost-saving has saved something on a session, you can see precisely what it saved. Open Token usage from the session menu and you get the lifetime totals — tokens saved, dollars saved, the share of your input tokens, and how many requests it covers — plus a chart you can read by hour, day, week, or month, and a breakdown of which model the savings came from. Everything shown is measured where your agent actually runs, never estimated by us.

Totals that survive restarts

Lifetime tokens saved, dollars saved, the share of input tokens compressed, and the request count — they keep counting across restarts, not just for the current run.

Read it at your own zoom

Switch the chart between hour, day, week, and month to tell a busy afternoon apart from a steady month.

Which model saved the most

A per-model breakdown shows where the savings concentrate — repeated context on one model usually saves far more than a model you barely use.

Export what you see

Export the visible window to CSV — including the model column — so you can drop the numbers into your own spreadsheet or expense report.

  1. 1
    Turn on cost-saving

    Token usage reports what cost-saving saved, so enable it on the session first.

  2. 2
    Let the agent work

    Savings are measured as your agent runs — a long session with repeated context has the most to show.

  3. 3
    Open Token usage

    From the session menu (⋯) on Home, Sessions, or inside the session, choose Token usage.

  4. 4
    Zoom, compare, export

    Switch between hour, day, week, and month, check the per-model split, and export the window as CSV.

NOTE

The Token usage row appears on a session once that session has reported savings — the same signal behind the green savings banner. How much you save depends on your own work: a young session may show a low single-digit share of input tokens, while a model with a lot of repeated context can reach roughly ten percent. Requires a recent CLI on the session — older sessions won’t show the row.

FAQ

Related

Cost-saving (token compression)Cloud (GitHub Codespaces)Self-hosted (your own server)

Ready to try it?

Start freeSee pricing