How API usage is billed
API requests are paid in two layers. First they use your plan’s included monthly tokens, at no extra charge. Once a plan window is full, they can continue on your prepaid API credit, if usage credits are turned on.
Layer 1: your plan allowance
Each paid plan includes a monthly token allowance shared across the organization’s usage, including API calls and Syntic Code. While all three windows (session, weekly, monthly) have room, requests carry no per-token charge. On the Console’s Billing Details page these show tokens with no cost.
| Plan | Monthly tokens |
|---|---|
| Starter | 2,000,000 |
| Business | 2,000,000 |
| Pro | 5,000,000 |
| Scale | 8,000,000 |
| Operations | 25,000,000 |
| Enterprise | Custom |
The Free plan (200,000 tokens) can’t use the API.
Layer 2: API credit
When a window is full and usage credits are on, requests are paid from the organization’s credit balance at your plan’s API rates. If credits are off, or the balance is empty, the request gets 402 quota_exhausted.
You add credit on the Console’s Recharge page: minimum $25, with bonus credit from $100. Auto-reload can top it up for you. The Billing overview shows the balance.
Keeping costs predictable
- Cap credit spending with the Monthly spend limit on the Billing overview. Credit-paid requests stop once it’s hit.
- Choose the lightest model that does the job. See Models and how to choose one.
- Set
max_tokensso replies can’t run longer than you need. - Reuse the same opening instructions across requests: repeated prompt beginnings are billed at the lower cached rate, shown in the Cached column of Billing Details.
Where to see what you spent
| Question | Page |
|---|---|
| How much of my plan is used? | Usage Limits |
| What did I spend per day or per model? | Billing Details |
| What credit did I add? | Recharge Details |
| Can I get it as a spreadsheet? | Export Records |