Aller au contenu principal
Velqa
← Back to home

Rate limits and budgets on Velqa.dev

Velqa applies per-key API limits to protect your balance and smooth out load: a maximum throughput (RPM / TPM) per model, and a monthly USD budget per key. These limits depend on your plan and the model; the exact values are shown in your dashboard.

The three limits

LimitWhat it measuresScope
RPMRequests per minuteper key × model
TPMTokens per minuteper key × model
Monthly budget (USD)Cumulative spend over the monthper key
  • RPM / TPM: short-term throughput ceilings. They depend on the plan and the model — a Premium model has lower ceilings than a Fast one. Exact values are visible in your dashboard, next to each key.
  • Monthly budget: a maximum spend amount per key that you set yourself. Useful to isolate a project or environment (prod / staging) with no risk of overrun.

How limits are enforced

  • Exceeding RPM or TPM → HTTP 429 (Too Many Requests). Retry after a short pause (exponential backoff), or spread the load across the minute.
  • The key's monthly budget exhausted, or the Boost balance insufficient → HTTP 402 (Payment Required). Top up, raise the key's budget, or wait for the monthly reset.

Plans (Starter, Dev, Pro) set global ceilings; Recharge Boost mode draws on a prepaid balance. See Available models for availability per plan.

Configuring per-key limits

You set the monthly budget and, where applicable, the throughput ceilings per key from the dashboard. See the API keys guide for how to do it. Best practice: one key per project/environment, with a dedicated budget.

Interaction with velqa-auto and Boost fallback

If you use the velqa-auto router (automatic model selection) and/or Boost fallback, exceeding a plan ceiling can automatically route the request to your Boost balance instead of failing, depending on your configuration. See Boost fallback for the exact behavior.

See also