Claude usage
5-hour resets 11-18 13:00 · weekly resets 11-21 10:00 · fixed demo sample
i
Input, output and cache tokens (read and write) of every request in the range.
i
Estimated API-equivalent cost at list prices as of 2026-10-25. Requests on a model without a price are left out and listed below.
i
Cache-read tokens as a share of all tokens. A high share means most context was served from the cache.
i
Model requests that carried usage counts.
By day
| Day | Tokens |
|---|---|
| 2026-11-12 | 6,096,286 |
| 2026-11-13 | 4,921,805 |
| 2026-11-14 | 3,281,739 |
| 2026-11-15 | 4,271,902 |
| 2026-11-16 | 7,149,797 |
| 2026-11-17 | 8,373,248 |
| 2026-11-18 | 11,251,999 |
Breakdown
| Org | Claude tokens | Share | Requests | Cost | Cursor sessions | Cursor agent-hours |
|---|---|---|---|---|---|---|
| Claude | Claude tokens45.3M | Share100% | Requests441 | Cost$37.30 | Cursor sess.- | Cursor h- |
Claude tokens come from the Claude Code session logs (any agent of the org that ran on Claude). Token usage is not reported by Cursor CLI: it leaves no token or usage counts on this machine, so Cursor runtime agents show sessions and agent-hours per model from the agents table, and no cost.
Attribution quality
| Matched by | Sessions |
|---|---|
| session | 16 |
| event | 0 |
| name | 35 |
| worktree | 0 |
| unattributed | 0 |
Costs are estimates at list prices as of 2026-10-25; a trailing + means some tokens were on models without a price.