Chargeback
The Platform Usage dashboard shows you which teams, workspaces, and users are consuming AI Units and Enterprise h2oGPTe tokens — and, for AI Units, which products — so you can attribute spend accurately and export it to CSV for chargeback in your own FinOps tooling.
Workspaces are the primary dimension for cost attribution. Teams or cost centers may own one or more workspaces, so usage can be aggregated across a team's workspaces for chargeback.
Prerequisites
- You must belong to the
telemetry-admingroup and be assigned theadminrole to access Platform Usage data.
Step 1: Select the billing period
- Open Platform Usage from the top-right dropdown.
- Click the date range button.
- Select MTD to cover the current month-to-date, or enter a custom start date and end date to match your billing period.
- Set Granularity to Monthly for a single aggregated total per workspace, or Daily for a time-series breakdown.
- Click Apply.
Step 2: Review AI Unit usage by Workspace
- Set Resource to AI Unit Usage.
- Set Group By to Workspace.
- Review the summary cards for totals and the stacked bar chart for each Workspace's share of consumption.
To filter to specific Workspaces:
- Open the Workspace entity filter and select the Workspaces you want.
- Select Apply.
Use the Metric Type selector to break the normalized AI Unit total down into its underlying requested resources (CPU, memory, or GPU). These reflect the amounts each pod requests (reserves), not real-time utilization.
Step 3: Review token usage by Workspace
If your organization uses Enterprise h2oGPTe and tracks LLM consumption separately:
- Set Resource to Enterprise h2oGPTe Token Usage.
- Set Group By to Workspace.
- Set the date range to match your billing period and select Apply.
- Review the Total Tokens, Input Tokens, and Output Tokens summary cards.
Step 4: Export the data
- Confirm your filters (resource type, date range, group by).
- Select the download icon above the data table.
For details about the CSV format and what each metric export includes, see CSV export in Platform Usage.
Step 5: Apply cost rates
The platform exports raw consumption values. Apply your organization's rates to the exported CSV in a spreadsheet or FinOps tool to calculate costs per Workspace.
Common scenarios
Monthly chargeback by team
- Set the date range to the first and last day of the billing month.
- Set Granularity to Monthly.
- Set Group By to Workspace.
- Select the download icon to export.
Per-user attribution within a Workspace
If a Workspace is shared and you need user-level breakdown:
- Set Group By to User.
- Open the Workspace entity filter and select the target Workspace.
- Select Apply, then export.
Quarterly comparison between two teams
- Set the date range to cover the quarter.
- Set Granularity to Monthly.
- Set Group By to Workspace.
- Open the Workspace filter, select the two Workspaces, and select Apply.
- Export the CSV. The file includes one column per month.
GPU requests by Workspace
- Set Resource to AI Unit Usage.
- Set Group By to Workspace.
- Set Metric Type to GPU.
- Export the CSV.
GPU values are based on Kubernetes pod requests (i.e., the amount of GPU capacity each pod reserves). A pod that requests 2 GPUs contributes 2 GPU-hours to the total regardless of how much of that capacity it actually uses.
Sharing a view
Dashboard settings persist in the browser URL, so you can copy the URL to share an exact view with colleagues. The recipient must have telemetry-admin access to see the data. For URL parameter reference and examples, see Sharing and bookmarking in Platform Usage.
- Entity filters reset when you change the date range. Reapply filters after changing dates.
- The platform exports raw consumption values. Cost calculations must be done outside the platform.
- Token usage covers Enterprise h2oGPTe only. Other LLM integrations are not tracked in Platform Usage.
Related topics
- Platform Usage — Full reference for the dashboard filters, controls, and visualizations.
- User Access — Manage Workspace membership and group assignments.
- Submit and view feedback for this page
- Send feedback about AI App Store to cloud-feedback@h2o.ai