Where to find it
- Click Control Tower in the sidebar.
- Choose a scope tab (Team, Workspace or Organization) and pick the entity in the selector.
- Set the date range.
The cost sections
Cost Over Time
Total spend per day across the selected period. For long ranges, the chart shows one bar per week. Use it to spot spikes, for example after a rollout or a new scheduled task.Cost by Model Over Time
The same spend as stacked bars, split by model. It shows when a model starts or stops being used.Cost by Provider
Spend grouped by AI provider (OpenAI, Anthropic, Google, Mistral and others), with each provider’s share, request count and cost.Cost by Model
Each model’s share of spend, request count and total cost. Switch between Cost by Model and Requests by Model, and between Chart view and Table view.Usage by Model
The detailed table: Requests, Prompt Tokens, Completion Tokens, Total Tokens and Cost per model.Reading the numbers
- All amounts are in euros, written in European format, for example
1.234,56 €. - Prompt Tokens are what is sent to the model: your message, earlier messages in the chat, attached files and instructions. Completion Tokens are what the model writes back.
- Avg Cost / Request is total spend divided by total requests.
- In a shared chat where several members write, each message counts toward the member who sent it.
- Usage can take a few minutes to appear.
Keep costs under control
- Check the top model in Insights. If an expensive model leads, check whether members need it for their tasks.
- Watch long chats. Every reply re-sends the earlier conversation, so prompt tokens grow with chat length. Starting a new chat for a new topic keeps costs down. See Context window.
- Review premium models. Decide whether your organization needs them. See Premium models.
- Find heavy users. Use Usage by User to find out who drives spend, then share good practices with them.
Export
Click Export CSV to download the usage for the current scope and date range.Related
Analytics
Scopes, date ranges and the full dashboard.
Usage limits
What happens when a limit is reached.
Model guide
Pick the right model for a task.
Model availability
Models used for built-in tasks.

