Account & admin
Insights
See where and why your usage goes: what costs the most, which steps are slow or fail, which models are efficient, and when your runs happen. Open Insights from the left rail.
The period
The dates at the top set the period every chart covers: one month ending now, unless you pick other dates. Refresh loads the latest; Last month brings back the default.
What the page shows
| Section | Charts |
|---|---|
| Where spend goes | Spend by model, by project, and the workflows that spend the most. |
| Heaviest steps | The most expensive steps, and the slowest, on average per run. Each step is named with its workflow. |
| Reliability | How many model requests succeeded, the error rate of each model, and the steps that fail most. |
| Efficiency | What each model costs per 1,000 tokens, and how its tokens split between input and output. |
| When | Requests by weekday and hour of the day, in your time zone, and requests by hour. |
Hover a chart for its numbers. Download on each chart saves its numbers as a CSV.
Totals over time are on Usage and spend; each step's own records are in The step console.
Something wrong or missing? Write to [email protected] about “Insights”.Updated 11 Oct 2026