Managing Credits

How pricing works

Every model is priced per million tokens, from roughly $0.05 per million tokens for small open-source models up to $10 per million tokens for the largest frontier models. Input and output tokens are typically priced separately (output usually costs more). Each request's usage object in the response reports exactly how many tokens were billed — see Making your first request. Check the model catalog in the dashboard for current per-model rates before committing to one in production.

Adding credits

Add credits from the billing page at app.aniron.ai. Credits apply account-wide, across every API key on the account — they are not scoped to a single key. Individual keys can still be capped with a per-key budget so that no single service or environment can spend the whole balance; see Best Practices.

Checking your balance

The usage dashboard shows your current balance, burn rate, and a per-model, per-key breakdown of spend. It's the source of truth for balance — build alerting or budget automation on top of the values it reports rather than estimating from your own token counts, since exact per-model pricing can vary.

Low-balance behavior

Once the balance reaches zero, new requests fail with 429 insufficient_credits — the same status code as a rate limit, but a distinct code value so you can tell the two apart programmatically:

{
  "error": {
    "message": "Account credit balance is depleted. Add credits to continue making requests.",
    "type": "rate_limit_error",
    "code": "insufficient_credits"
  }
}

A depleted balance is not something a fallbacks array can route around — every model draws from the same account balance. Set a low-balance alert threshold in the dashboard so you can top up before requests start failing.