BIGGEST RECORDED DAY
—
—
THE TOKEN LEDGER
A little perspective
on a lot of tokens.
A running record of my AI usage.
Two providers, over time.
Loading usage history…
OpenAI account-reported tokens + retained local Anthropic tokens. Coverage differs between sources.
COMBINED RECORDED TOKENS MIXED COVERAGE
—
Loading…
LONGEST STREAK
— days
—
01 / ACTIVITY
The bigger picture
—tokens across recorded history
Recorded tokens—
02 / RHYTHM
A day at a time
Hover or focus a day to inspect it. No record Less More
Dates are shown as returned by OpenAI.
TOKEN COMPOSITION
What makes up the total
Total = uncached input + cache creation + cache reads + output. Cached tokens count toward volume; this chart does not estimate spending.
How closely does this match actual usage?
By model
03 / BREAKDOWN
By the numbers
| Month | OpenAI | Anthropic | Tokens | Share |
|---|---|---|---|---|
| Total recorded | — | — | — | 100% |
Select a month to explore its daily activity. First and latest months may be partial.
04 / THE FINE PRINT
Real numbers.
Clear boundaries.
01
Reported by the account
These are token counts returned by OpenAI’s account usage endpoint, not estimates from conversation length.
02
Coverage has a boundary
Classic ChatGPT chat coverage is not verified. This source does not separate ChatGPT from Codex, or break totals down by model, input, output, or cache.
03
Read the source documentation ↗Anthropic source documentation ↗A snapshot, with a timestamp
Loading snapshot details…