What these tokens would have cost billed per token at each model's published price. It is the one figure that compares across people on different plans.
List-rate estimate / day · last 90 days
Jun 23, 2026peak $817.09Sep 20, 2026
on a plan$14,951
paid per token$2,011
per active day$176.69
31.42B
tokens
327.3M on a working day
163,029
requests
192.7K each
96
active days
not a day missed
96
day streak
still running
Activity
+22% vs previous 30 days
31.42B tokens over 96 active days · last 12 months
Averaging 327.3M per working day, with a median of 290.0M. Busiest was Sep 16, 2026 at 1.34B. Cells are shaded by what the day was worth at list rates, not by token count. Hover a day for the models and tools behind it.
less spendmore spend
Last 12 months in full on this profile, last synced Sep 20, 2026. Counters only: no prompts, no project names, no file paths.
Drift
What got reached for, month by month
Each row is one month scaled to its own 100%, so a quiet month and a loud one take the same width and only the mix changes.
Output is a fraction of the tokens but usually most of the money: it is priced at several times the rate of cache reads.
cache read
$11,913
cache write
$2,210
output
$2,112
fresh input
$727.95
Every model here has a known rate.
Cache
Cache does most of the reading
A raw token total mostly measures re-read context. The ratios below show how the work was done.
98.4%
of everything entering the context window came from cache instead of being paid for again
264.3×
tokens of context read per token written
192.7K
tokens in an average request
cache read
97% · 30.44B
cache write
1% · 383.7M
output
0% · 118.4M
fresh input
2% · 481.3M
Models
68 models, mostly claude-opus-5
claude-opus-5
9.89B · 31.5%
$6,077 estimated at list rates · 36,713 requests
claude-opus-4-8
4.28B · 13.6%
$3,248 estimated at list rates · 15,448 requests
muse-spark-1.3-contributor
3.72B · 11.8%
$23.74 estimated at list rates · 20,824 requests
deepseek-v4-flash
1.98B · 6.3%
$30.81 estimated at list rates · 7,988 requests
claude-fable-5
1.62B · 5.2%
$2,727 estimated at list rates · 9,414 requests
grok-4.6
1.47B · 4.7%
$913.77 estimated at list rates · 12,228 requests
claude-sonnet-5
1.20B · 3.8%
$453.27 estimated at list rates · 10,546 requests
muse-spark-1.3-contributor-free
1.19B · 3.8%
$7.84 estimated at list rates · 6,053 requests
gpt-6-astra
1.02B · 3.3%
$1,480 estimated at list rates · 7,655 requests
gpt-5.6-sol
697.8M · 2.2%
$414.05 estimated at list rates · 5,660 requests
+58 more below one percent.
Tools
17 tools, mostly Claude Code
Claude Code
17.91B · 83,009 requests · $13,135
57%
Muse
4.30B · 24,860 requests · $140.14
13.7%
Codex
3.74B · 22,248 requests · $1,986
11.9%
Grok Build
2.05B · 16,117 requests · $1,138
6.5%
OpenCode
2.05B · 12,619 requests · $29.49
6.5%
Cursor
1.10B · 1,696 requests · $433.80
3.5%
Antigravity
174.4M · 1,795 requests · $59.17
0.6%
Hermes Agent
59.6M · 34 requests · $37
0.2%
Qwen Code
33.3M · 450 requests · $2.08
0.1%
DeepSeek Harness
3.0M · 48 requests · $0.06
0%
Kimi Code
1.5M · 31 requests · $0.95
0%
Devin CLI
758.2K · 15 requests
0%
Cline
392.2K · 80 requests · $0.14
0%
OpenClaw
230.1K · 11 requests · $0.05
0%
Kilo Code
196.3K · 10 requests
0%
Pi
34.8K · 5 requests · $0.06
0%
Copilot CLI
17.0K · 1 requests · $0.01
0%
Rhythm
Heaviest on Wednesday
Tokens by day of the week, in the timezone each device reported.
Mon: 4.12B over 13 days
Tue: 4.09B over 13 days
Wed: 6.30B over 14 days
Thu: 5.78B over 14 days
Fri: 4.58B over 14 days
Sat: 3.17B over 14 days
Sun: 3.39B over 14 days
MTWTFSS
By month
4 months tracked, busiest Sep 26.
Jun 26: 3.17B over 14 days
Jul 26: 7.75B over 31 days
Aug 26: 10.07B over 31 days
Sep 26: 10.44B over 20 days
JunJulAugSep
Hour of day is not in the sync payload yet, so the rhythm here stops at the day. See docs/tokenstat-sync-cli.md for what the CLI would need to send.
Spread
Across 127 projects
Project names stay on the device. What reaches us is an opaque hash we can count but never read, so this page shows how the work was spread, never which work it was.
project 1
49.4%
project 2
10.8%
project 3
9%
project 4
7.7%
project 5
6.8%
project 6
3.5%
project 7
3.4%
project 8
2.9%
Ordered by size, unlabelled on purpose. The top project is 49.4% of everything.
How the counts were read
Some tools report token counts exactly; the rest are derived from what they write to disk. The chart keeps that split visible, so an inferred number never poses as a measured one.
