Budget control
LLM budget control
When several teams share models, spend drifts fast. Tokenjoy passes quota down from the company to members and API keys, checks it on every call, and blocks calls when it runs out. No waiting for the monthly bill.
Quota flows down level by level
- Company budget: the overall usage budget
- Department / team budget: allocate quota and roll up usage
- Project / member quota: set per business purpose
- API key: tied to quota, permissions and expiry
Blocked when spent, not reconciled afterwards
Each call is checked in order: key validity, requested model, company account status, wallet balance, key status and expiry, department and key model permissions, and finally the available quota of the key and its member or project. When quota is gone the call is rejected. Calls that pass are billed by actual usage, without double counting.
Alerts and top-up approval
- Budgets follow the company time zone and calendar month, with usage alerts sent through configured channels
- When quota runs out, a member or project can request more, and keep working once approved
- Extra quota is limited by the balance that can still be allocated
The wallet settles, budgets allocate
The company wallet handles settlement and organization budgets handle allocation, managed separately. Cost trends and token usage can be broken down by department, member, model and channel.
FAQ
How do I set a spending limit per team on an LLM API?
Allocate a budget to each department, then to projects, members and API keys. Every call is checked against the available quota and rejected when it is used up.