> ## Documentation Index
> Fetch the complete documentation index at: https://docs.noxus.ai/llms.txt
> Use this file to discover all available pages before exploring further.

# Billing analytics

> What the usage and cost figures under Settings → Billing → Analytics mean

**Settings → Billing → Analytics** shows how your organization used AI Operations and AI models over a time range. Pick **Billing cycle** to see the same period that limits and charges use.

## AI Operations used

Every node run, chat message and ingested file counts as one AI Operation. This is the number your plan's AI Operations allowance and your usage controls count, so it matches **Settings → Billing** for the same period.

## Recorded AI Operations overage

What the organization was charged for AI Operations beyond what the plan includes, in the selected time range.

* It always covers the **whole organization**. Selecting a workspace does not split it, because one charge can come from usage in several workspaces.
* Each charge keeps the price that applied when it was made. Changing the plan or a limit later does not rewrite past charges.
* It shows in the currency of your plan. If charges were made in two currencies, both totals are shown.

The figure is zero when overage is turned off or usage stayed within the plan.

## AI model usage cost

What your AI model calls cost in the selected time range. When part of it is estimated, the amount has an asterisk. Hover over it to see how the total is made up:

| Line | What it means |
| - | - |
| **Reported by the model providers** | The cost of calls whose provider reported the tokens used. |
| **Estimate for calls still running or incomplete** | Calls that have not finished yet, or never reported their usage. They are counted at their expected cost until the provider reports. |
| **Estimate for calls without a usage report** | Finished calls whose provider sent no usage. They are counted at their expected cost. |
| **Total** | The sum of the lines above. Model usage budgets count this same amount. |
| **Not counted: interrupted or retried calls** | Calls that stopped before an answer, for example because a run was cancelled or the call was retried. They are shown for reference and are not part of the total. |

A short note under the breakdown compares the estimates in the total with what the providers reported, and warns when many calls were interrupted or retried, which can cost more than shown.

Model providers charge in USD. If your plan is in another currency, such as EUR or GBP, amounts are converted at the European Central Bank rate for the start of the billing period, so they do not change during the period. When no rate is available, amounts stay in USD and the tooltip says so.

<Note>
  Health checks that test a model connection are never counted.
</Note>

## Workspaces

With **All workspaces** selected, the totals include usage from workspaces that were deleted, because it still counts toward the organization's limits. The charts per workspace and per tool then show a **Deleted workspaces** bar, so the bars add up to the totals. The bar only appears when there is such usage.

## Why a figure can differ from Settings → Billing

* **Time range.** Settings → Billing always shows the current billing cycle. Analytics shows the range you pick.
* **Workspace filter.** Usage figures follow the selected workspace. Recorded overage always covers the whole organization.
* **Calls still running.** Estimates for running calls are replaced by the reported cost when each call finishes.


This documentation is built and hosted on [Mintlify](https://mintlify.com), a developer documentation platform.