Activity by weekday and hour
1 week to Tue 27 Oct 2026, 12:28, local time UTC-04:00. Shade is uncached input plus output tokens for codex.
| Weekday | 0 | 1 | 2 | 3 | 4 | 5 | 6 | 7 | 8 | 9 | 10 | 11 | 12 | 13 | 14 | 15 | 16 | 17 | 18 | 19 | 20 | 21 | 22 | 23 |
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Mon | ||||||||||||||||||||||||
| Tue | ||||||||||||||||||||||||
| Wed | ||||||||||||||||||||||||
| Thu | ||||||||||||||||||||||||
| Fri | ||||||||||||||||||||||||
| Sat | ||||||||||||||||||||||||
| Sun |
Hover or focus a cell for its figures.
Tokens from21,23086,648187,990335,559517,034
Model by task kind
1 week to Tue 27 Oct 2026, 12:28, codex only. Requests by canonical model and the session's task label; a subagent counts under its parent's label. Column totals match the label totals of query summary --group-by label --since 1w (codex rows).
Each web scales its largest topic to the outer ring. Ring ticks and axis labels give the share of that model's requests.
gpt-6-astra
184 requests OpenAI
gpt-6.1-sol
2,715 requests OpenAI
gpt-6-sol
6 requests OpenAI
gpt-6-luna
407 requests OpenAI
Exact request counts by task kind
| Model | research | coding | debugging | writing | review | data-analysis | ops-config | other | Total |
|---|---|---|---|---|---|---|---|---|---|
| gpt-6.1-sol | 174 | 306 | 152 | 13 | 908 | 112 | 1,001 | 49 | 2,715 |
| gpt-6-luna | 127 | 6 | 10 | 0 | 60 | 0 | 182 | 22 | 407 |
| gpt-6-astra | 31 | 68 | 0 | 31 | 34 | 0 | 0 | 20 | 184 |
| gpt-6-sol | 0 | 0 | 0 | 0 | 6 | 0 | 0 | 0 | 6 |
| Total | 332 | 380 | 162 | 44 | 1,008 | 112 | 1,183 | 91 | 3,312 |
Requests from62034112182
How much of my work I hand to subagents
1 week to Tue 27 Oct 2026, 12:28, codex only. Tokens are uncached input plus output. Figures match query agents --since 1w (codex row of by_tool).
Share of tokens spent in subagent threads vs main threads. Each square is 5 %. By month covers all history.
| Row | Squares | tokens | sessions with subagents |
|---|---|---|---|
| codex | 53 % | 2 of 21 |
| Row | Squares | tokens |
|---|---|---|
| May | 0 % | |
| Jun | 4 % | |
| Jul | 2 % | |
| Aug | 10 % | |
| Sep | 15 % | |
| Oct, so far | 45 % |
subagent threadsmain threads
Cost by project
1 week to Tue 27 Oct 2026, 12:28, codex only. Sessions, tokens and cost match query projects --since 1w counting codex only.
| Project | Sessions | Tokens | Est. cost | Tools |
|---|---|---|---|---|
| AsterX | 106 | 335,972,224 | $109 | 100 % |
| design | 4 | 44,244,018 | $10 | 100 % |
| t3code | 3 | 34,606,435 | $5 | 100 % |
| project R | 2 | 18,976,246 | $3 | 100 % |
| aweSOM | 8 | 4,302,550 | $2 | 100 % |
| project H | 1 | 206,297 | $0 | 100 % |
| project BO | 1 | 163,632 | $0 | 100 % |
Linked projects are public on GitHub. Other project names are pseudonyms shared across this export. Tool shares are percentages of requests.
What my use would cost at API prices, by week
1 week to Tue 27 Oct 2026, 12:28, codex only. Estimated cost by week, counted back from now, at API list prices; the total row matches query summary --since 1w. Figures are rounded independently to the nearest dollar; displayed rows may not add up.
| Week | codex |
|---|---|
| from 20 Oct | $129 |
| Total | $129 |
96 % of input tokens were cache reads, priced at the cheaper cache rate.
How often I exhaust a window limit
1 week to Tue 27 Oct 2026, 12:28, codex only. Closed windows of each plan's own limits; pools are left out.
5 hr
Each bar is every closed window, by how full it got before it reset. A tool needs 5 closed windows for a bar.