Usage
Request volume, errors and latency — and the two things Tessen cannot yet tell you.
Your own request volume, error rate, latency percentiles and per-key activity are real and readable. The endpoint is scoped to the caller; there is no user parameter.
GET /api/devstats →Cost
Tessen does not report cost. The column exists in the usage table, every measured live row is zero, and no pricing for model consumption is defined anywhere in the product. The Console renders an em dash rather than a zero amount, because a rendered zero reads as "we measured your spend and it was nothing" — which is a claim, not an absence.
Intelligence units
There is no normalized unit of Tessen intelligence, and raw provider token counts are not one. Nothing in this documentation converts an execution into units, and no multiplier or rate is published, because none is defined. Token counts are shown where they exist, labelled as the technical consumption they are.