generated: '2026-09-07'
method: searched
source: https://docs.toapis.com/docs/cn/api-reference/rate-limits/async-tasks.md
limit_count: 6
notes: >-
  Published limits cover async image/video task submission and status queries only;
  two independent token buckets that do not consume synchronous model-request quota.
  Limits are counted per user ID, not per token — all tokens owned by one user share
  the bucket. Business failures count; requests rejected with 429 do not. A batch
  status query counts as one request (max 100 task IDs). Synchronous chat/completion
  rate limits are not published. Rate-limit groups (default/vip/svip) are account
  tiers assigned by the provider.
limits:
  - scope: per-user async task submission (group default)
    limit: 300 RPM
    sustained: 5 RPS
    burst: 50 per 10s
  - scope: per-user async task submission (group vip)
    limit: 1000 RPM
    sustained: 16.7 RPS
    burst: 167 per 10s
  - scope: per-user async task submission (group svip)
    limit: 2000 RPM
    sustained: 33.3 RPS
    burst: 334 per 10s
  - scope: per-user async task status query (group default)
    limit: 3000 RPM
    sustained: 50 RPS
    burst: 500 per 10s
  - scope: per-user async task status query (group vip)
    limit: 10000 RPM
    sustained: 166.7 RPS
    burst: 1667 per 10s
  - scope: per-user async task status query (group svip)
    limit: 30000 RPM
    sustained: 500 RPS
    burst: 5000 per 10s
headers:
  - Retry-After
  - X-RateLimit-Limit
  - X-RateLimit-Remaining
  - X-RateLimit-Reset
  - 'X-RateLimit-Category: async_submit|async_query'
exhaustion:
  status: 429
  error_code: rate_limit_exceeded
  degraded:
    status: 503
    error_code: rate_limit_unavailable
    note: returned when the submission rate-limit service is unavailable; queries degrade to pass-through
guidance: >-
  Provider recommends webhooks as the primary completion signal with polling as
  fallback: poll at 5-10s minimum with jitter, honor Retry-After on 429/503, use
  exponential backoff when the header is absent.