generated: '2026-08-12' method: searched source: >- https://open.kuaishou.com/docs/develop/server/errorCode-new.html and https://open.kuaishou.com/docs/develop/IndustrySolutions/introduction/errorCodes.html (both HTTP 200, server-rendered, fetched 2026-08-12); response headers observed on https://open.kuaishou.com/openapi/user_info and https://api-singapore.klingai.com/v1/videos/text2video (2026-08-12) notes: >- Kuaishou publishes throttling as ERROR CODES rather than as a rate-limit reference page, and returns no rate-limit response headers on any surface probed. That is the finding: an integrator can learn that they were throttled only by parsing the body, and cannot learn their remaining budget at all. Two concrete numeric limits ARE published, both for local-services product mounting. Everything else is undocumented. limit_count: 2 signal_headers: published: [] observed: [] note: >- No X-RateLimit-*, RateLimit-* or Retry-After header was returned by any Kuaishou, Kwai or Kling host. The only correlation headers present are X-REQUEST-ID and X-KSLOGID on open.kuaishou.com. On Kling AI the correlation value is carried in the response body as request_id, not in a header. apis: - api: kuaishou:open-platform exhaustion: - code: 10000302 message: api too fast title: Generic throttle http_status: 200 envelope_field: result detail: >- Documented as "请求频率太快被限流" (request frequency too high, throttled). No numeric threshold, window or scope is published for this code. source: /docs/develop/server/errorCode-new.html - code: 10002032 message: 操作过于频繁 title: Per-product operation throttle http_status: 200 envelope_field: result source: /docs/develop/IndustrySolutions/introduction/errorCodes.html limits: - id: product-mount scope: per external product id + poiId operation: 商品挂载 (product mount) limit: 1 window: 10 minutes burst: null error_code: 10002032 note: >- The provider states the limits verbatim: "当前限制:商品挂载10min 1次,商品编辑10s 1次". Limits for the same product under different POIs are independent of one another. source: /docs/develop/IndustrySolutions/introduction/errorCodes.html - id: product-edit scope: per external product id + poiId operation: 商品编辑 (product edit) limit: 1 window: 10 seconds burst: null error_code: 10002032 source: /docs/develop/IndustrySolutions/introduction/errorCodes.html non_api_quotas: - id: merchant-withdrawal scope: per merchant per channel limit: 1 window: 1 day error_code: 10000645 note: A business quota surfaced as an API error, not a request-rate limit. source: /docs/develop/server/errorCode-new.html - api: kuaishou:kwai-for-business limits: [] note: >- No rate-limit documentation is reachable on the public surface; developers.kwai.com is a JavaScript single-page application and its reference is gated behind advertiser registration. - api: kuaishou:kling-ai limits: [] concurrency_note: >- Kling AI is an asynchronous generation API — work is submitted, then polled by generationId — so the meaningful constraint is concurrent-task and credit budget rather than requests per second. Neither a published concurrency table nor a rate-limit header was found. The provider's MCP/CLI guide does publish one hard constraint on the agent surface: MCP and CLI accept only PAID credits from the Personal workspace, and bonus credits cannot be spent through them. exhaustion: - code: 1001 http_status: 401 title: Missing Authorization note: Observed; not a throttle, recorded here only because it is the only live code confirmed. source: https://kling.ai/app/mcp/guide gaps: - >- No numeric request-rate limit is published for the Kuaishou Open API generally, for the Kwai for Business marketing API, or for the Kling AI generation API. - >- No rate-limit response headers are emitted anywhere, so an agent cannot read remaining budget or a reset time from a live response — it can only detect the throttle after the fact from result code 10000302 or 10002032.