How Do I Understand My Teams Token Usage As An Admin
As an admin in Brim, you can export a detailed CSV of your organization's token usage directly from the Token Usage Dashboard. This export gives you a row-by-row breakdown of compute consumption across all projects, models, and variables — useful for billing reconciliation, reporting to stakeholders, or digging into usage patterns outside the dashboard.
How to export token usage
- Go to Settings
- Click Token Usage
- Apply any filters you want (project, date range)
- Click the Export button to download the CSV
What's in the export
Each row represents token usage for a specific project, model, and variable combination within a given month. Here's what each column contains:
|
Column |
Description |
| Project Name | The name of the project |
| Project ID | The unique ID for the project |
| Month | The billing period in YYYY-MM format (e.g., 2026-05) |
| Model Name | The LLM model used (e.g., gpt-4o-mini) |
| Variable | The variable that triggered the generation |
| API Calls | Number of API calls made |
| Tokens Used | Total tokens consumed (input + output, excluding cached) |
| Input Tokens | Tokens sent to the model as input |
| Output Tokens | Tokens returned by the model |
| Cached Input Tokens | Input tokens served from cache (not counted in Tokens Used) |
| Created Date | When the record was first created |
| Last Updated | When the record was last updated |
Understanding token types
The export breaks token usage into three separate columns : Input, Output, and Cached Input so you can see exactly where compute is being spent.
Input tokens are the tokens sent to the model: your prompt, variable context, and any document content passed in. These tend to be the largest share of usage.
Output tokens are the tokens the model generates in response. Output tokens are typically more expensive per token than input tokens with most LLM providers, so this column is useful for cost calculations.
Cached input tokens are input tokens that were served from your provider's cache rather than reprocessed. Brim excludes these from the Tokens Used total to give you a cleaner picture of actual compute consumed. If you ran the same generation twice in quick succession, you may see the second run's cached tokens here.
Using the Month column
The Month column uses YYYY-MM format (e.g., 2026-04 , 2026-05 ). This format sorts correctly in Excel and other spreadsheet tools without any additional formatting just sort the column alphabetically to get chronological order.
This makes it straightforward to: - Pivot by month to compare billing periods - Filter to a specific month or range - Group by project and month to see trends
Tips for working with the export
- Filtering before export: Apply date range and project filters in the dashboard before exporting this limits the export to only the data you need.
- Note on data availability: Token usage data is only available from the date your institution upgraded to the December 2025 version of Brim. Usage before that date is not reflected in the export.
To learn more about where token usage is visible within the Brim app, you can find that information here.