1,110 GitHub contributions alongside 20.4B AI tokens
1,110
SAME DATES · ACTIVITY, NOT CAUSATION
› LOCAL HISTORY / CLEAR NUMBERS
Your AI coding, measured.
Scan Claude Code, Codex, OpenCode, Command Code and Hermes logs in your browser. See tokens, speed, models, rhythm and GitHub activity.
SAMPLE / 30 DAYS
Fictional history. Real components. Your scan supplies your numbers.
Total tokens
20.4B
99% is cached context your tools re-read
20,430,723,847 tokens through 12 models · 12 named
Cache read20.2B98.7%
Input158M0.8%
Cache write28.7M0.1%
Output76.3M0.4%
API value
$3,378
LIST-PRICE ESTIMATE · 98% OF REQUESTS PRICED
New tokens
263M
INPUT + OUTPUT + CACHE WRITE
Models
12
12 NAMED · SAMPLE
Current streak · all time
23days
LONGEST 23 DAYS · ALL TIME
Sessions
3,079
29 DAYS ACTIVE
23 days in a row with AI in this period
23 days
CURRENT AND LONGEST RUN THIS PERIOD
Cached context would have cost $26,330 more at full input price
$26,330
MATCHED CACHE READS AT FULL INPUT PRICE
DeepSeek-V4.1-Flash accounted for 50% of your tokens
50%
10.4B OF 20.4B TOKENS
01
Tokens in, code out
Tokens above the line. Public GitHub contributions below.
TOKENS / DAYGITHUB / DAY / SAMPLE
GitHub contributions
1,110
ILLUSTRATIVE PUBLIC CALENDAR
Tokens / contribution
18.4M
COUNTS ACTIVITY, NOT CODE QUALITY
02
Speed
Output tokens per second. Band shows the middle half of replies.
MODEL04080120160MEDIANWAITREPLIES
0 tok/s160 tok/s
Gemini 3.5 Flash
120.8tok/s
5.0s
1,836
GLM 5.3 Flash
110.0tok/s
5.4s
2,222
Claude Haiku 4.5
103.1tok/s
5.7s
1,017
DeepSeek-V4.1-Flash
96.1tok/s
6.3s
64,063
GPT-6 Astra
78.7tok/s
7.7s
4,523
GPT-6.1 Sol
67.8tok/s
8.4s
263
Claude Sonnet 5.5
67.2tok/s
9.0s
3,229
GPT-6 Sol
62.2tok/s
9.7s
18,510
GLM 5.3
52.7tok/s
11.5s
9,128
DeepSeek-V4-Pro-0813
51.0tok/s
11.9s
3,643
Gemini 3.1 Pro
43.9tok/s
13.8s
5,539
Claude Opus 5.5
38.0tok/s
15.9s
12,310
Fastest median
120.8tok/s
Gemini 3.5 Flash
Replies timed
126,283
FROM PROMPT TO FINISHED REPLY
03
Models
Where the tokens went, and their value at API list prices.
| MODEL | DEVELOPER | TOKENS | SHARE | API VALUE | TOK/S |
|---|---|---|---|---|---|
| DeepSeek-V4.1-Flash | DeepSeek | 10.4B | 50.7% | $68 | 96.1 |
| GPT-6 Sol | OpenAI | 2.98B | 14.6% | $748 | 62.2 |
| Claude Opus 5.5 | Anthropic | 2.02B | 9.9% | $711 | 38.0 |
| GLM 5.3 | Z.ai | 1.47B | 7.2% | $417 | 52.7 |
| Gemini 3.1 Pro | 891M | 4.4% | $309 | 43.9 |
04
Rhythm
Requests by weekday and hour. Sample times are UTC.
MON
TUE
WED
THU
FRI
SAT
SUN
0003060912151821
MODEL CALLS · 0 … 4,122
Peak hour
21:00
UTC
Busiest day
SEP 25
BY REQUESTS
After midnight
10%
MIDNIGHT TO 5 AM
Weekends
25%
06
Private by design
Your logs never leave this browser. Sharing uploads only the numbers on your card.
- READ
- Model names, token counts and timestamps from the history files you choose.
- NEVER READ
- Prompts, responses or code. No sign-in files or account profiles.
- RUNS HERE
- A Web Worker in your browser counts your history on this device.
- WHAT LEAVES
- Nothing from your logs. A share link uploads only your card's numbers, GitHub gets only your username, and the site counts page visits with Cloudflare Web Analytics.
07
Your tools. One view.
Supported local coding histories, together.
Claude CodeCodexOpenCodeCommand CodeHermesT3 Code
Models ↗83
Published prices and exact identities
Providers ↗29Model developers and access routes
Benchmarks ↗44Reported scores with original evidence