One employee's spend for the period: the two line items attributable to a person — agents (by agent) and chat (by model), each against its own weekly cap. Opened by clicking a row in the AI usage list. Cost in dollars, a daily trend and export — v2.
Employee detail
employee · this week
achilles.local/admin/usage/42
/admin/usage/:user · Owner / Admin
Dmitry Sokolovdmitry@company.com
Period · this week ▾
1.20M / 2M
Agents · tokens this week 1
60% of cap · blocking · all agents
340k / 500k
Chat · tokens this week 2
68% of reference · non-blocking
1.54M
Total
agents + chat
Chat · by model 4
| Model | Messages | Tokens |
|---|
| GPT-4o |
168 |
250k |
| Claude Sonnet |
42 |
90k |
—
Cost v2
input/output price by model
—
Daily trend v2
spend chart for the period
-
1
Agents against the weekly cap — a blocking one: past it, runs wait for the week to reset.
Agent Engine.
-
2
Chat against the reference cap — an "used against the limit" indicator that does not block requests (v1).
-
3
The agent breakdown is by agent, not by model: the run journal
(
agent_runs.tokens_used)
does not store the model per run, so the "Model" column shows the agent's current model; a precise by-model breakdown of agents — v2.
-
4
The chat breakdown is by model: each reply records the model
(
messages.model) and tokens
(messages.tokens_used), so the breakdown is exact.
-
Cost in dollars, a daily trend and export — v2.