Usage, costs, and activity details
Steinkauz AI shows usage and costs for your active organization across web chat and API Access in one place. Switch in the organization switcher to see a different organization’s usage.
Tokens remain telemetry, not invoice truth. The Tools tab lists generic tool_invocations records (including client-defined Completions tools), not a Steinkauz AI tool storefront.
Where to find it
Open Usage & Activity from the main navigation. The page has three tabs:
| Tab | What it shows |
|---|---|
| Cost | Spend over time, optional Budget remaining when you set caps, breakdowns by model or provider |
| Inference | Searchable log of model activity from chat and API requests |
| Tools | Client-reported tool calls from chat or API completions (name, status, timing, attributed cost when known) |
Filter by date range and, on the Inference tab, by source (Chat or API), model, API key, provider, and trigger type.
In organizations, owners and admins can also filter by member (Myself, All members, or a specific person) to audit costs and inference across the organization. Admins see metadata only for other members (costs, tokens, models, timing; not titles, prompts, or message content). Owners can open full activity details including prompts/snapshots and Open chat for audit. Members always see their own usage.
Active organization
Usage, budgets, and billing actions always reflect your active organization: the organization’s Cloud or Private Deployment entitlement, and optional Budgets in Settings → Budgets.
Organization owners see Manage billing for the active organization from this page. Only owners can open the billing portal for an organization. See Organization billing.
Chat and API in one view
Web chat and API Access share the same activity log for the active organization. Each row is one user-facing request. Source badges distinguish Chat from API (with API key label when applicable).
Customer-owned inference
Chat and API draw from each Budget subject’s optional caps on each provider instance. A customer-owned Gateway connection still needs your Gateway API key.
When Remaining reaches zero for a configured Budget unit, new requests for that subject on that provider are blocked until you raise the cap or the period resets. See Usage Budgets.
Steinkauz AI does not bill inference. Usage is still recorded so you can reconcile with provider accounts. BYOK stays uncapped by Steinkauz AI until you set optional Budgets.
What you can inspect
Click any row in the Inference or Tools tab for activity details:
- Model and provider per step
- Tokens and cost per step and for the full request
- Duration and time to first token
- Policy decision recorded for the request
- Client-reported tool calls when present (inputs, outputs, and attributed cost when known)
- Message context and JSON export (owners for other members’ rows; always for your own)
Multi-step requests appear as one entry with multiple steps.
Activity details in chat
- Session activity: Chat top bar → activity panel for the current conversation; link to Usage & Activity filtered to that session.
- Message activity: Activity button on assistant replies for that response’s details.
Cost and budget summary
| Organization | |
|---|---|
| Budget | Optional caps per member / dedicated key in Settings → Budgets |
| Exhausted | Block chat/API when Remaining ≤ 0 for that subject × provider |
| Inference | Customer-owned (BYOK / customer Gateway) |
For plan details, see Plans overview. For who changed organization settings (including routing blocks and sensitivity events), see Configuration Audit.