specification: API Commons Rate Limits
specificationVersion: '0.1'
schema: https://raw.githubusercontent.com/api-evangelist/interface-research/main/schema/api-commons.yml#/$defs/RateLimits
provider: Kling AI
providerId: kling-ai
created: '2026-07-11'
modified: '2026-07-11'
reconciled: false
tags:
- Video Generation
- AI Video
- Generative AI
- Rate Limiting
- Concurrency
description: >-
  Kling AI's generative work is asynchronous, so throughput is governed less by
  a per-minute request cap and more by how many generation tasks an account may
  run concurrently. Concurrency allowances scale with account tier and prepaid
  spend; higher tiers and enterprise agreements raise the number of simultaneous
  in-flight video/image tasks. The lightweight create and query (polling) calls
  are subject to request-rate limiting; exceeding it returns HTTP 429. Kling
  does not publish a single authoritative numeric table, so the figures here are
  indicative and NOT reconciled.
notes: >-
  Concurrency and request-rate numbers vary by account tier and are not
  authoritatively published. Reconcile against the Kling AI Open Platform
  console and documentation. Poll query endpoints on a backoff (e.g. every few
  seconds) rather than tightly, and honor 429 responses.
sources:
- https://app.klingai.com/global/dev/document-api/apiReference/commonInfo
- https://app.klingai.com/global/dev/document-api/quickStart/productIntroduction/overview
responseCodes:
  throttled: 429
limits:
- name: Concurrent Generation Tasks
  scope: account
  metric: tasks
  limit: tier-based (not authoritatively published)
  notes: Number of simultaneous in-flight video/image tasks; grows with account tier and prepaid spend.
- name: Task Creation Requests
  scope: account
  metric: requests
  limit: not published
  notes: Rate on POST create-task calls; 429 on exceed.
- name: Task Query (Poll) Requests
  scope: account
  metric: requests
  limit: not published
  notes: Rate on GET query-task calls used for polling; 429 on exceed.
- name: JWT Token Lifetime
  scope: token
  metric: seconds
  limit: ~1800 (about 30 minutes)
  notes: Tokens are short-lived; regenerate and cache rather than reusing indefinitely.
policies:
- name: Async Task Model
  description: Submit a task, receive a task_id, then poll the matching query endpoint until status is succeed or failed. Assets return as URLs that expire and should be downloaded promptly.
- name: Callback Notifications
  description: A callback_url may be supplied on task creation to be notified on completion instead of tight polling.
- name: Backoff Strategy
  description: Poll on an interval with exponential backoff and jitter; honor 429 responses and any Retry-After header.
maintainers:
- FN: Kin Lane
  email: kin@apievangelist.com