Logs lists individual inference requests, newest first — the requests behind the totals on Usage. Every request is one row; there is no grouping or aggregation.
Filters
The header sets the scope: range, member (when your role allows it), and a date range. The toolbar narrows further:
- A text box that matches a full request id, or part of a model’s slug or name.
- A model filter and a status filter, each accepting several values.
All filters live in the URL, so a filtered view can be bookmarked or shared. Changing any filter returns to the first page.
Rows are always sorted by time, newest first. The table cannot be sorted by another column, because the API does not accept a sort parameter.
Columns
| Column | Content |
|---|---|
| Member | Shown when the scope is not just you. |
| Model | The model slug, with the owning account below it. |
| Path | The request path; streaming requests are labelled. |
| Tokens | Hover for the input, cache, and output breakdown. |
| Cost | Three decimal places; hover for the breakdown. |
| Latency | Two lines: time to first token and total time. |
| Status | Success, failure, or cancellation, shown by icon and colour. |
| Time | When the request arrived. |
Latency figures exist only for requests that reported them. When the gateway reports no cost or usage for a request, the row says so instead of showing zero.
Scope and permissions
- A model owner can see their own requests or those of all their models.
- An admin can see their own requests and those of the members they manage.
- A member sees only their own requests.
The server clips the scope before any filter applies, so filters only narrow what you may see.