generated: '2026-09-07' method: searched source: https://doc.advance.ai/liveness_detection.html docs: https://doc.advance.ai/liveness_detection.html summary: >- ADVANCE.AI documents rate limiting and quota behaviour but publishes no numbers. The Glossary describes the signals an over-limit response carries — an envelope code plus a Retry-After HTTP header — without stating any request-per-second, per-minute or daily figure, and without naming the scope the limit applies to. Limits are account-specific and negotiated with a sales manager, so an integrator cannot size a client from the public documentation. limit_count: 0 limits: [] signals: headers: - name: Retry-After returned_on: [SERVICE_BUSY, RETRY_LATER] documented_example: 'Retry-After: 10s' note: >- Returned as an HTTP header on a 200 response whose envelope code is SERVICE_BUSY or RETRY_LATER. This is the only runtime rate-limit signal ADVANCE.AI publishes. - name: X-RateLimit-* present: false - name: RateLimit-* present: false status_code_on_exhaustion: 200 status_code_note: >- Exhaustion is signalled in the envelope, not the HTTP status. ADVANCE.AI states SERVICE_BUSY "may migrate to HTTP 429 Too Many Requests in the future" — a client should handle 429 today as well, but must not rely on it. envelope_codes: - code: SERVICE_BUSY billed: false meaning: 'Rate limit is exceeded, please retry after the suggested time in HTTP Header.' also_used_for: 'Quota exceeded: You have exceeded the daily quota for free queries.' action: Honour Retry-After, then back off exponentially. - code: RETRY_LATER billed: false meaning: Query failed, please retry after the suggested time in HTTP Header. action: Honour Retry-After. - code: OVER_QUERY_LIMIT billed: false meaning: >- Free query quota exhausted, for a production account or for a test account. Distinct from SERVICE_BUSY — this is an entitlement ceiling, not a throughput limit. action: Not retryable. Contact the ADVANCE.AI sales manager for a paid allocation. - code: INSUFFICIENT_BALANCE billed: false meaning: Prepaid balance exhausted. action: Not retryable. Top up. quotas: free_daily_quota: exists: true value: null note: >- A "daily quota for free queries" is named in the SERVICE_BUSY description and a separate free query quota is named in OVER_QUERY_LIMIT, including a distinct test-account allocation. No figure is published for any of them. scope: documented: false note: >- Whether limits apply per access key, per account, per endpoint or per country is not stated. IAM_FAILED shows entitlement is scoped per country and per calling domain, which suggests limits are account-scoped, but ADVANCE.AI does not say so. gaps: - No published numeric limit for any operation. - No X-RateLimit-* or RFC 9238 RateLimit-* headers, so a client cannot see remaining budget before exhausting it. - No published quota figure for the free tier or the test account. - Retry-After is documented in prose with one example value; its exact format (seconds vs HTTP-date) is not specified.