Skip to main content
Antara bills usage in USD from your organization’s prepaid credit balance. Each model has a price per million input, cached input, and output tokens (pricing in GET /v1/models). Reasoning tokens are output tokens.

​
Adding credits

In the console’s Billing page you can:
  • Buy credits as you go: $20, $100, $500, or a custom amount.
  • Subscribe to a plan (Developer, Growth, or Scale). Each plan’s monthly price comes back as the same amount of credit, and higher plans grant a higher rate-limit tier.
  • Turn on auto-reload, which buys credits when the balance drops below a threshold.
Credits don’t expire. Prices and plans are listed by GET /v1/plans (no authentication needed).

​
How a request is charged

  1. Admission. The request needs an available balance above zero, and spend under any spend limit. Antara reserves the request’s estimated cost.
  2. Settlement. When it finishes, the actual tokens are charged and the reservation is released.
Because admission only needs a positive balance, one request can take the balance a little below zero; the next one then gets 402 insufficient_credits.

​
Spend limits

Set a monthly cap on the organization or on a project (UTC calendar month). When month-to-date spend reaches it, requests get 402 spend_limit_reached until the next month or until you raise it. Manage limits in the console or with the spend limit endpoints and an Admin API key with spend_limits:write.

​
Checking usage