Share by installs
Share by installs
Cumulative · Share of installation counts across listed agents. Installations using multiple agents are counted more than once.
AGENTCAT / TRENDS
A working day.
With a few more cats.
Which agents do we work with? Which cats keep us company? Explore the patterns in AgentCat’s anonymous aggregates.
AGENTCAT / DATA
Loading public aggregates.
01 / Agents
The agents we work with
Compare the share of installations with the share of tokens processed.
Estimated token share
Estimated token share
Local month counters from installs reporting within 26 hours · Processed tokens include cached input, not billed tokens.
02 / Models
The models in use
The models on the most installs in the last 7 days.
Top 5 by installs
Top 5 by installs
Last 7 days · Share of installation counts within these five models. Multi-model installations can appear more than once.
03 / Mascots
The company we keep
Who spends the most time alongside us? A look at activity recorded with each mascot.
The company we keep
The company we keep
Cumulative · Share of activity events within these five mascots, rather than a ranking by users.
04 / Activity
The rhythm of work
Busy days, working hours, and where usage is concentrated.
Activity by weekday
Activity by weekday
Last 30 days · KST · Each day’s share of all activity events.
Activity by hour
Activity by hour
Share of published hourly activity. All bars use the same percentage scale.
Highest daily usage
Highest daily usage
Previous completed KST day · Observed increases in cumulative tokens. No user identity is shown.
How usage is distributed
How usage is distributed
Local month counters received within the last 26 hours · Token share of the top 10% of installations.
05 / Quota weather
Unofficial quota weather
Aggregate quota signals and external reset forecasts, with their sources.
CODEX / COMMUNITY SIGNAL
Unofficial quota weather
How to read these numbers
These are AgentCat’s public anonymous aggregates. Installation counts use anonymous installation IDs. Small samples are withheld, and missing data is never shown as zero. Users of multiple agents or models can be counted in more than one category.
Tokens measure processing, including cached input, not billing. Lifetime counts growth since each install’s first retained observation, excluding pre-Agent Cat history. Outliers are capped; with at least 20 installs, the top 1% is clipped to the 99th percentile.
Provider token shares and usage distribution use local month counters from installs reporting within 26 hours. Week/month figures are lower bounds for those reporting installs. Install and mascot counts are cumulative; models cover 7 days, weekdays and hours 30 days. Rounding may prevent a 100% sum; update times vary by metric.
View public data ↗ Data policy ↗