Usage

Your organization's usage charges, who is driving them, where the rest of the month is heading, and what the work produced.

The Usage page lives at /admin/usage and requires the View usage & billing permission (canViewUsage) — a sensitive permission carrying the "Owner-reserved" badge (Strata's escalation ceiling; there is no separate Owner role — built-in roles are admin and user). The built-in admin role holds it by default, so admins see this page out of the box. Drill-downs into individual users open at /admin/usage/<userId>.

Note

Usage covers consumption — usage charges from your credit ledger, requests, and token volume. Your subscription itself (plan, seats, invoices, payment method) lives on the separate Billing page, which the same canViewUsage permission unlocks read-only; changing the plan additionally requires canManageBilling.

Requests for more usage

When a member has asked for a higher monthly spend limit, their request appears at the very top of the page; the panel is absent when nobody is waiting. Strata does not send admins a notice when a request is filed, so this panel is where requests are found. Admins with Manage billing also find it on the Billing page, because that permission answers a request but does not by itself open this one. Each request shows who asked, what they had used of their limit when they asked, their reason and, if their limit has changed since, what it is now. An admin with Manage billing enters the new monthly limit and chooses Approve, or chooses Decline; an admin who can only view the page is told who can act instead. The new limit must be above the one that applies to the member now (the field suggests twice that), and it becomes their own limit until someone changes it; a lower or equal amount is refused and the request keeps waiting. Either way the member is emailed the answer, and it appears in their Strata inbox. A member whose request was declined can ask again whenever they need to. A request filed because the member's team budget had run out says which group's budget, with Edit group limits: raising that budget answers it. See Spend limits for members.

The control bar

A sticky bar at the top of the page sets the window for every figure below it — the summary, the chart, the tables, time saved and key insights — except the compute-credits card, which pages by calendar month on its own. It was previously inside one chart's header, which read as though it only drove that chart.

• 7 days / 30 days / 90 days pills, with the window's actual dates printed beside them (for example Aug 23 – Sep 22); there is no free-form date picker. The pill row shows whichever window is active rather than sitting with nothing selected. Longer windows — 180 and 365 days — are reachable from the Key insights dropdown.
• Forecast toggle. Projects the calendar month's usage charges: this month's charges already inside the window, plus the rest of the month at the window-average daily rate, so it is anchored on real charges rather than a pure extrapolation. The window average divides by the days that actually have data, not the full range. The 30-day and longer windows always reach back to the 1st. On the 7-day window, from the 9th of the month on, only the last seven days count toward the month-to-date part, so the projection can come in low. It appears as a line under the headline — "$X projected this month, at $Y/day" — and as faded bars on the By surface chart. Neither appears when the window has no usage charges.

The summary

Usage charges is the dominant figure — delivered work from your credit ledger, priced at $0.50 a credit — with the credit count beneath it ("12 credits at $0.50 each"). While credits are not being invoiced, that line ends "· accrued, not yet invoiced". Questions answered, People using Strata ("N of M", counting people only — API service accounts are left out) and Agent runs (with agents' share of usage charges beside it, when they drew any) are the supporting numbers.

A further figure appears only while someone is actually being throttled — it is absent rather than zero the rest of the time, which for most organizations is most of the time. It reads "People are at their hourly limit", or "Person is at their hourly limit" when it is exactly one, above a Review link. Throttled people are shown in red in the Members table: "Users in red are currently at their hourly limit."

Usage charges, Questions answered and People using Strata each carry a period-over-period comparison that names the prior absolute value — for example "↑ 2127% from $5.50 in the previous 30 days". It reads new when the previous window was zero, flat under a 1% move, and is left off entirely when both windows are zero. Agent runs carries no comparison.

Note

These comparisons are deliberately colourless. Higher spend is not a failure and lower spend is not a win: a drop can mean the team stopped using what you are paying for. The number tells you what changed; only you know whether that is good.

Where the money goes

Usage charges over time is a single trend chart with two views:

• By surface — stacks each day's usage charges by where the work came from. Chat and Agents always show. API (API keys), Outlook (the add-in), Teams and Voice join only when they drew credits in the window. The support assistant draws no credits, so it never appears here. Hover a day for the breakdown.
• By model — daily token volume, split by the model that produced it: the top six, with the rest folded into Other. Chat-side calls only, and volume rather than money — a credit is earned by a deliverable, not by a model, so there is no per-model charge to plot. Bar and line rendering apply to this view.

The time axis plots every day in the window, including days with no activity, so a quiet week is visible as a gap rather than being closed up. Once today has any activity, the chart's caption adds "Today is still counting" — today is a part day, and undecorated it reads as a collapse.

Below the chart, the Usage by model table breaks the same window down per model, by volume.

Usage by model (detail)

Between the chart and the Members table is a model-level breakdown — one row per model your organization has actually used in the window, ranked by token volume. Each row carries the model's name, provider, and logo, with Uses and Tokens columns. There is no dollar column: usage charges come from deliverables, not from models. The first 12 rows show, with a Show N more models button for the rest, and the export button beside the title downloads the table as CSV (model, provider, uses, tokens). The rows are whichever models from your Model Catalog generated activity, so the list reflects the models you have enabled rather than a fixed lineup.

Note: scheduled-agent runs are not in this table — an agent's runs and credits roll up under its owner in the Members table. The table covers your organization's other model usage — regular chat plus non-chat auxiliary AI such as Knowledge Base embeddings, the vision document-review chain (PDF/DOCX/XLSX/PPTX), image OCR, the in-app support assistant, the Strata for Outlook add-in, and the assist helpers.

Feature labels

Auxiliary AI calls are labeled by feature with a model-id suffix, so platform helpers are distinguishable from regular chat:

The assist helpers carry no suffix, so they appear under their model's plain name. They include chat-title generation, agent-capability suggestions, writing-voice distill/rewrite, announcement draft/edit, memory import, and analytics categorization. They run on the cheapest small chat model your organization has enabled (for example gpt-5.4-mini or another economical model from your catalog), so the row they appear under reflects your org's own model set.

The Strata for Outlook add-in spans several base models. Rather than scattering them, all -outlook rows are collapsed into a single Strata for Outlook row, whose uses and tokens add up the add-in's turns across every model they ran on. Sub-agents started from the add-in are not folded in: they count on that model's own (Sub-agents) row.

People and seats

Members table

Titled Members and ranked by usage charges in the window. If your organization has any API service accounts or connection bots, including ones left behind by revoked keys or removed bots, a Members / API accounts / All switch sits beside the search field; it starts on Members, and the heading, the first column and the count under the table follow it.

An API account is listed under its key's name — "Nightly ETL +1" when it has more than one key (hover for every name) — and described as "API account · last used 3d ago", never by an internal address. While one of its keys works it carries no status word. When none does it reads Expired when every key has expired, and Revoked otherwise — for each key, the word the API keys page uses, where a key that was revoked reads Revoked whatever its expiry date — and Retired once the account itself has been retired. A connection bot outlives its keys, so it only ever reads Retired. API accounts with no working key and no use in the selected window are held back: a line under the table counts them — "2 API accounts with no working key had no use in this range and are hidden." — beside Show them, which lists them (and becomes Hide them) for the rest of your browser session. An account with use in the window is always listed, because its spend is in the totals.

Columns:

• Member (API account or Account, following the switch) — display name + email; for an API account, its key's name and when it was last used.
• Groups — the person's groups; shown only when your organization uses groups and your role can see them.
• Chat queries — count of chat requests.
• Agent runs — count of scheduled agent runs.
• Credits — credits recorded against them in the window; credits from an agent's runs count toward the agent's owner, and an API call's toward the actor its API key is bound to (a team member, a connection bot, or the key's own service account).
• Usage charges — those credits at $0.50 each, with a sort arrow on this column by default.

The table shows 100 rows at a time, with Show N more for the next; a search, a sort, the switch and Show them each start it again from the top, and rows with the same value come most recently active first. Every column header except Groups is sortable. Users currently at their hourly limit have their name rendered in red (hover for how much of their cap they have used). A search field above the table filters by name, email, key name or group. The export button to the right of search downloads exactly the rows in view as CSV — the search, the Members / API accounts / All switch and Show them all apply, every page of the table is included, not just the rows showing, and the rows keep the table's sort order — with totals for the selected window. The columns are Name, Email (empty for an API account), Role (Admin or Member, as the row says it; API account or API bot for one), Groups (when your organization uses groups), Active, Status (Expired, Revoked, Retired or Disabled; empty otherwise), Last active (for an API account, when it was last used), Chat queries, Agent runs, Credits and Usage charges ($). The column headings, yes and no, the role and the status words are in your language; numbers and dates stay the way a spreadsheet reads them (12.50, 2026-09-15). The file starts with a byte-order mark, so Excel opens it as UTF-8 and a heading in Japanese or a name like Zoë arrives intact; the export of the models table does too.

Click any row to open /admin/usage/<userId> — the per-user billing profile.

Per-user billing profile

The profile page shows every metric for one user in detail:

• Their usage charges over the window, beside chat queries, agent runs and their credit count for the current calendar month (UTC). While weighted metering is off, that last tile reads Billable actions · metering not enabled.
• A daily chart of their queries, and a By day table of queries, tokens and errors that you can filter to days with activity or with errors.
• Their token volume by model, and up to 20 of the scheduled agents they own, with each one's model, status, runs and tokens.
• Key insights on their pattern: how many times a usage limit stopped them in the window (their hourly limit, or a spending, credit or token-pool cap), and, if they are at their hourly cap right now, how much of it they have used this hour.
• Their conversations, when your organization has member-chat oversight turned on and your role can view member chats. A transcript opens read-only, and each opening is recorded in the audit log. Select rows to export them: as a ZIP archive, whose download link is emailed to you, works only for you and expires 24 hours after it is sent, or as a CSV or JSON file of up to 500 conversations that downloads in your browser.

Use this when investigating a heavy user, building a chargeback report, or troubleshooting why someone's agents run more than expected.

Seats

Below the Members table, a Seats card sets what your subscription bills for against who actually used Strata in the window: a bar split into used (active in the window), idle (has an account, no activity) and unassigned (paid for, nobody in them), with the share of billed seats in use. It counts people only: an API account holds no seat, so API traffic never fills one. It appears once your organization has a subscription; if the billed seat quantity and the count admission is measured against differ, the card names both.

Spend limits

Spend limits lists every group with its member limit, what each of its members may use in a month, and its shared budget, one monthly amount its members share (see Shared budgets for groups). The shared budget column shows what has been used of it this month, with a bar that turns amber at 75% and red at 90%. A dash means no shared budget, Paused a budget that is kept but not enforced, and Frozen a $0 budget. The first 25 groups are listed, with Show all. When the table is wider than the room it has, on a phone or on a laptop where the Organization page's list of sections takes its share, each group is shown as a card that names each figure (Members, Member limit, Shared budget and, with a custom order, Pays). Anyone who can view this page sees the table; Edit limits and the two settings above it need Manage billing.

Above the table, Member limit from groups decides which group's member limit applies to a member in several groups with limits: Lower group limit (the default) or Higher group limit. A limit set on the member in Users wins over both. Which shared budget pays first is Largest first (the default) or Custom order: with a custom order a Pays column numbers the groups that have a budget, starting from the order they pay in now, and the arrows beside each number move it up or down. The first budget in the order with money left pays.

Edit limits sets both together. Under Shared monthly budget choose No shared budget, Set amount or, for a budget that exists, Paused; under Member monthly limit choose No member limit or Member limit. Amounts are suggested from last month: about three times what the group's members used, for the budget, and about three times its top spender's use, for the member limit (Use $…). Choosing a budget brings a member limit with it, and choosing No member limit removes the budget, because a shared budget needs one. If the member limit is too low for the budget ever to run out, the dialog says so and offers to raise it. Saving a budget that has money left answers the requests it caused.

A View group button in a budget alert opens this page on the group's row.

What you're getting for it

Time saved, compute credits and key insights sit together at the foot of the page.

Time saved is modeled, not measured, and the card says so: the deliverable counts are real records, but the minutes attached to each kind are Strata's standard estimates per deliverable: 30 per document, 60 per dashboard, 20 per analysis, 10 per export, 5 per turn of actions carried out, and 15 per completed unattended agent run. Chat answers count as zero. The subtitle lists whichever assumptions are in play for the rows shown, so it reads differently depending on the work your organization did, and the hours are also given as working days of 7.5 hours. Treat the hours as "what our stated assumptions imply", not as an observation — it is the one figure on this page that is not a direct record of something that happened.

Compute credits

Below Time saved is a Compute credits card that tracks credit activity one calendar month at a time; it does not follow the date window set in the control bar. Credits are Strata's unit for finished work — agent runs, generated documents, Analysis runs, dashboards, and actions with an outside effect (batched once per request) — counted by the piece rather than by token. Today each of these records a flat one credit; size weighting applies only once weighted metering is switched on (see How usage is counted). While credits are not invoiced, the card's subtitle says so: "Recorded for reconciliation — not currently invoiced." Each credit is priced at a flat $0.50 on every plan; see Billing → Compute credits.

A month pager at the top of the card moves between months (UTC calendar months, back to July 2026, when the ledger began). The card shows:

• Credits recorded — the headline total for the selected month, beside its dollar value at $0.50 a credit. While weighted metering is off, that value is labeled "Indicative value (metering off — counts, not weights)".
• By type — the total split across the kinds of activity that draw credits (agent runs, documents, automated tasks, analyses, data exports, dashboards), with credits and event counts. A map or place lookup that Strata pays a vendor for is listed too, as Third-party calls, with its count and no credits.
• By user — a User / Credits table of up to eight people with the most credits in the selected month, highest first.

Note

The card always reads by calendar month, but during a free trial the allowance Strata enforces is the trial's 150 credits over the 14-day trial window — so a trial that crosses a month boundary shows its credits split across two months here. Once the 150 are spent, Strata stops accepting new requests for the rest of the trial, ordinary chat included, and tells members the trial has used its full allowance.

How usage is counted

This describes how credit activity is counted, not what you are charged:

Note

Size-weighted counting is how credits are recorded once weighted metering is switched on for your environment. Until then, credit events record flat — one per event — and the card labels the total indicative rather than billable ("metering off — counts, not weights"). If your totals read like plain event counts rather than the weights below, that's why.

Once weighted metering is on, credits are fractional and scale with the size of what Strata actually produced:

The rule underneath the table: credit follows output, not effort. A question that takes several tool calls and minutes of reading to answer well draws nothing; only a deliverable draws credit. Batching works in your favour either way — flagging forty emails in one request is a single credit event, not forty.

Two ceilings bound a runaway deliverable so a single request can never draw an arbitrary amount:

• No single deliverable draws more than 20 credits ($10). That is exactly the weight of a 200-slide deck, a 100-page report or a 10,000-row export; anything larger is clamped to 20, so nothing past the ceiling is counted.
• Analysis (code execution) draws at most 5 credits per chat turn, however many executions the turn takes — a ceiling that also applies while metering is off. The turn is the unit you perceive; the individual tool rounds are an implementation detail.

These do not draw credit:

• Work that failed, was cut off by a spend cap, or was rejected in review.
• Strata's own quality-control pass. Generated PDFs, Word files and complex workbooks are rendered and inspected, and the regeneration that fixes a broken one draws no credits. The inspection is itself model work: it runs in your own Azure AI Foundry like any other model call, so Microsoft bills you for that inference directly.
• Re-running the same task after a failure.
• A SQL question answered inline in chat. Persisting it — a saved dashboard, a large export — is a deliverable and does draw credit.

Note

The compute-credits card is a usage view — it measures credit activity, it is not a bill. Your subscription lives on the separate Billing page.

Key insights

A summary block at the foot of the page, beneath the compute-credits card and just above the page's closing footnote — who has the highest usage charges, the most-used model, the peak day, the 7-day trend, who hits a limit most often, the error rate and more. An API account is named here the way its row names it, after its key. Computed from the same dataset as the rest of the page — picking a longer range from the dropdown there reloads the page at the new window.

How costs are calculated

Behind the page, Strata estimates a dollar cost for each model call from its token counts. You don't see those dollars on the Usage page. The monthly AI spend ceiling and the agent spend caps are enforced against them (see the end of this section). A member's or group's Monthly spend limit doesn't use them; it counts usage charges from the credit ledger. Rates come live from your organization's Model Catalog — the input/output price per 1M tokens recorded on each model — so when Kronisys updates a model's pricing, new estimates follow. A static table mirroring the catalog is used only during the brief cold-boot window before the live catalog loads. Prompt caching is priced in: cache reads count at the model's Cached input rate where the catalog lists one, and at the full input rate where it doesn't (and during that cold-boot window); Anthropic cache writes count at 1.25 times the input rate. Each figure is priced once, when the usage is recorded; activity recorded before cache-aware pricing is not re-priced.

The rates are visible from the i button beside the hourly rate-limit fields — a role's Rate limit and your organization's default rate limit. The AI model costs modal lists the models your organization has enabled with their input and output token prices.

These are estimates, and your Microsoft invoice can differ because:

• Batch-API discounts — 50% off for batched requests.
• Long-context tier surcharges — extra cost for requests in the larger context windows.
• Negotiated EA/MCA rates — your Microsoft subscription may carry custom rates.

These cost estimates feed the monthly AI spend ceiling and the agent spend caps. The Usage page itself shows your organization usage charges from the credit ledger, and its footnote reads: "These are USAGE CHARGES — delivered work recorded in your credit ledger, priced at $0.50 per credit. They do not include your per-seat subscription, and token counts shown elsewhere on this page are volume, not money."

Relationship to your Foundry billing

AI inference runs in your organization's own Azure AI Foundry, in your own Azure tenant.

Nothing on this page is your Microsoft bill. Model inference is billed by Microsoft directly to your own Azure subscription, at Microsoft's rates (see Bring your own Azure AI Foundry); the dollar figure comes from your Azure invoice. What this page shows of that inference is token volume, not money: the By model view and the Usage by model table cover chat queries only, and each person's profile breaks their tokens down by model and by scheduled agent, so you can attribute volume to people in your org. Usage charges are separate again: delivered work recorded in Strata's own credit ledger, not part of your seats and not part of Microsoft's bill. While credit metering is off for your environment, the summary marks them accrued, not yet invoiced.

What can go wrong

See Rate limits and quotas for the cap that produces the "at limit" indicator, and Bring your own Azure AI Foundry for the billing relationship with your own Foundry.

Related

• Rate limits and quotas
• Bring your own Azure AI Foundry
• Roles and permissions