Know the spend
Know what every
session costs.
Every session reports its context size, output tokens, and cache behavior. Watch spend live — on desktop or from your phone.
Know the spend
Every session reports its context size, output tokens, and cache behavior. Watch spend live — on desktop or from your phone.
Context, output, and cache activity attributed to the session that spent them.
Numbers update as the agent works, not after the invoice.
Check a long run from anywhere and decide whether it continues.
Token counts help explain how a session is using model context. Beeplane reads usage recorded by supported local agent sessions and presents the numbers alongside the work, so you can compare runs in context.
Input or context tokens describe what the model receives; output tokens describe what it generates. They answer different questions, so review both.
Claude session records can include cache reads and writes. Beeplane shows those counts when the source session reports them.
A large count may reflect a long task or a large context. Compare similar work and inspect the transcript before deciding what to change.
Beeplane can estimate cost from token counts and built-in rates. Provider billing can differ by model, plan, and current pricing, so use the provider’s usage page for the invoice total.
Context size, output tokens, and cache activity per session, as reported by the session.
Yes — the session list carries usage per run, so heavy hitters stand out.
No — usage is read from session records, not intercepted traffic.
Free to download. Works with Claude Code, Codex, OpenCode, and other agents on macOS or Linux.