Admin Settings → Model Pricing is your rate card. IntraLLM records how much each model is used; the prices you set here convert that into the monetary figures shown in Usage and credit calculations.
Why configure pricing
- Meaningful reporting — usage in tokens is technical; spend in currency is what budgets understand.
- Chargeback and showback — allocate AI cost to departments by group.
- Behaviour steering — when teams see that one model costs 10× another, they choose sensibly.
- Capacity justification — evidence for hardware or licence decisions.
Setting rates
| Step | Action |
|---|---|
| 1 | Open Admin Settings → Model Pricing. |
| 2 | For each model, enter the rates your organisation uses — typically per-unit token prices, and where relevant per-request or per-image/per-video rates. Model names can be listed individually or by pattern (for example a family prefix). |
| 3 | Save. New usage is costed at the new rates. |
Note:
For models you host yourself, "price" is a policy number, not an invoice — set it to reflect your internal cost of compute, or to intentionally discourage overuse of expensive models. Common approach: base rates on comparable external API prices, then adjust for your priorities.
Keeping the rate card honest
- Update when models change — new models appear with no price until you set one; unpriced usage can read as free.
- Version your decisions — note why a rate changed; rates become policy arguments.
- Review quarterly alongside Usage trends.