Spend and models
Two pages in the organization sidebar answer two questions. Spend answers "what are we spending on AI, and who is spending it?" in money. Models answers "which models does our work run on, and who may pick them?". Your own provider keys and local models connect the two: work on them never uses up anyone's usage limits.
In this section
| Task | Guide |
|---|---|
| Choose how work is routed, which models each team may pick, and see prices | The Models page |
| See what work will take before it starts, and what it took | Quotes and charges |
| Read AI spend and set budgets for teams and people | Budgets and spend |
| Run work on your organization's or your own provider key | Bring your own model |
| Use models on your computer or your own servers | Local models and self-hosted servers |
Who sees what
Admins open Spend and Models from the organization sidebar (Global Admins) or the workspace sidebar (other admins). Everyone else sees their own usage and spend in Profile → Usage, and picks a model for a chat from the model menu next to the message box, from the models their team allows.
Spend, usage limits and billing
- Usage limits cover Harmony Agent work on Harmony's models. Each person sees them as a share of a five-hour and a weekly limit. See Usage limits.
- Spend covers money: your own keys, extra usage, AI subscriptions and other AI tools.
- Billing turns both into one monthly invoice. See The Usage & billing page.