Understand Usage, Costs, and Limits
Reasoning reports usage for each session and shows the estimated cost when pricing is available. Organization budgets and provider limits can stop or delay new requests.
Read session usage
The session header shows context use, tokens, and cost. Select it to open View context usage for the breakdown, including the Session cost. Treat cost as an estimate; final billing belongs to the provider or your AI Cockpit organization.
Subagent usage is included in the parent session's total.
Check your budget
Open Settings → Profile to see the active organization and the current AI Cockpit budget available to your account.
Get a cost alert
Open Settings → Auto-Approve and set Session Cost Alert to an amount in whole US dollars. Reasoning shows a non-blocking alert when a session exceeds it. Leave the field empty to turn the alert off.
The alert does not enforce a spending limit. Your organization's budget policy and the provider's controls still apply.
Respond to a limit
When a request is blocked by a budget, payment, or provider limit, Chat explains which one and what to check. See Understand Chat errors.
- Confirm the active organization in Settings → Profile.
- Review the budget available to that account.
- Confirm that the selected model and provider are still available.
- Wait for the provider's limit to reset, choose another permitted model, or contact whoever manages your budget.
Starting a new session does not bypass an account, organization, or provider limit.
To choose cheaper models for subagents, quick tasks, and compaction, see Compare Models (pre-release).