What the data says
observed 30-day total
723.6B
723,583,346,387 tokens across returned daily rows
average per observed day
24.1B
30 of 30 daily observations returned (100% coverage)
peak completed day
65.2B
Jul 30, 2026
week over week
+33.7%
Latest seven completed days versus the preceding seven
The latest row matches the dataset generation date and may be incomplete. It remains visible in the chart and table, but is excluded from the peak and week-over-week comparison.
Daily observed tokens and trailing 7-observation average
Bars are daily totals. The amber ticks are the average of that day and up to six preceding observations.
Daily tokens measure activity observed among participating public developers. They are useful for spotting cadence and spikes inside this cohort, but a higher line does not by itself prove that AI coding usage is growing across the developer population.
This page requests the public insights endpoint with period=30d and sorts up to 30 valid daily rows by date. Negative, non-finite, or malformed observations are excluded.
The trend field is always a trailing 30-day series, independent of other period controls. Totals on this page are recomputed from the displayed daily rows so the chart, table, and headline remain internally consistent.
The newest returned date remains visible but is excluded from peak and week-over-week comparisons because submitter-local day boundaries make completeness impossible to prove from the aggregate response.
Participants opt into public listing, and the cohort can change as developers join, leave, or submit new aggregates. Changes can reflect cohort composition, reporting cadence, weekends, releases, or unusually large sessions—not only changes in individual behavior.
Dataset generated .
Aggregated model, agent, trend, skill, and tool-call totals from participating developers.
Explains what the CLI aggregates, what it does not collect, and the limits of the public cohort.