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

# Set budgets and billing ownership

> Assign future usage to the intended resource team and interpret budgets, reuse and pending prices.

You need resource Owner access to change its billing team. Use the appropriate administration or instance controls for budget settings.

## Inspect the recorded payer

Open the workspace or dataset settings. Check its billing team before starting substantial work.

| Resource or work | Attribution |
| - | - |
| New workspace | The creator's first-joined team. |
| New dataset inside a workspace | The creating workspace's billing team. |
| Work started from a workspace | The workspace's recorded team. |
| Work on the dataset itself | The dataset's recorded team. |
| Dataset enrichment binding | The dataset's team and the binding principal. |
| Ingestion run | Team recorded at admission from the destination dataset. |
| Personal Arc or Kato chat | Recorded user, without a workspace team on the chat. |

A collection or source has no separate billing team. Attaching or detaching a dataset does not transfer its payer.

## Change future ownership

As Owner, select a billing team that you belong to and save. Inspect the saved setting and a subsequent work record.

The team need not be a member of the workspace. Billing ownership grants no content access. Earlier usage retains its recorded attribution.

When a resource has no billing team, the default team pays. The default team cannot be renamed, assigned members or deleted. An Owner selects another team to move future spend.

## Set and check limits

For an app instance, set its spending limit per run in settings. Check what the limit covers and test a small run.

The app model-spend cap is a soft check between blocks. A block already running can exceed it. Storage and compute need their own applicable controls.

Inspect the root task's allowance, attempts and child work. Revisions and retries consume remaining allowance. A displayed context percentage measures context occupancy, not total cost.

## Review actual usage

Use Console usage views for the intended user, team, model and workspace. Match the reporting period to the work records.

Check inference, allocated compute and retained storage separately. Retried or failed work can have incurred usage. Reusing an existing result adds no new inference call, but retained bytes continue to use storage.

A missing price does not mean no usage. Keep measured quantities and pending prices visible until pricing or reconciliation completes. Estimates remain distinct from actual usage and invoices.

Review retained evidence obligations before reducing storage. See [Understand run outcomes and costs](/vantage/run-apps-and-workflows/run-outcomes-costs) and [Retention, deletion and legal holds](/administration/configuration-and-usage/retention-deletion-holds).


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