generated: '2026-08-26'
method: searched
source: https://docs.mthreads.com/kuaecloud/kuaecloud-doc-online/coding_plan/plan_overview
name: Moore Threads Published Limits
description: >-
  Moore Threads publishes real, numeric limits — but almost none of them are rate limits in the
  conventional sense, and NONE of them are signalled in a response header. The KUAE Cloud Coding
  Plan publishes a per-tier prompt budget on a 5-hour rolling window; the AIBook speech APIs
  publish payload and pacing ceilings instead of request-per-second quotas, and signal exhaustion
  only through the in-band numeric status codes 2003 and 2004.
limit_count: 13
headers:
  request: []
  response: []
  retry_after: false
  note: >-
    No RateLimit-*, X-RateLimit-*, or Retry-After header is documented on any Moore Threads
    surface. An agent cannot discover its remaining budget at runtime; it can only observe a
    failure after the fact. This is the single largest runtime-semantics gap on this provider.
exhaustion:
  http_status: 200
  in_band_codes:
  - code: 2003
    text: too many requests
    meaning: Concurrency limit exceeded
  - code: 2004
    text: service overload
    meaning: Client is sending audio faster than realtime
  - code: 3001
    text: service is busy
    meaning: Server-side saturation
limits:
- id: coding-plan-free-trial
  api: KUAE Cloud Coding Plan API
  scope: per-account
  window: 5 hours (rolling)
  limit: ~40 prompts
  burst: null
  note: >-
    Moore Threads benchmarks this as "equivalent to a Claude Pro subscription". Requests may queue
    during platform peak. Valid 30 days from claim; one claim per account.
  source: https://docs.mthreads.com/kuaecloud/kuaecloud-doc-online/coding_plan/plan_overview
- id: coding-plan-lite
  api: KUAE Cloud Coding Plan API
  scope: per-account
  window: 5 hours (rolling)
  limit: ~120 prompts
  note: Stated as 3x Claude Pro. Requests may queue at peak.
  source: https://docs.mthreads.com/kuaecloud/kuaecloud-doc-online/coding_plan/plan_overview
- id: coding-plan-pro
  api: KUAE Cloud Coding Plan API
  scope: per-account
  window: 5 hours (rolling)
  limit: ~600 prompts
  note: Stated as 3x Claude Max (5x). Higher generation speed and a response-latency guarantee.
  source: https://docs.mthreads.com/kuaecloud/kuaecloud-doc-online/coding_plan/plan_overview
- id: coding-plan-max
  api: KUAE Cloud Coding Plan API
  scope: per-account
  window: 5 hours (rolling)
  limit: ~2400 prompts
  note: Stated as 3x Claude Max (20x). Priority during peak. Docs note one prompt consumes roughly 15-20 model calls, and that the figure is an estimate that varies with project size.
  source: https://docs.mthreads.com/kuaecloud/kuaecloud-doc-online/coding_plan/plan_overview
- id: tts-text-length
  api: Moore Threads AIBook Text-to-Speech API
  scope: per-request
  window: null
  limit: 300 characters
  note: UTF-8; English words must be space separated. Applies to both the unary and streaming surfaces.
  source: https://docs.mthreads.com/tts/unary-tts/introduction
- id: asr-audio-size
  api: Moore Threads AIBook Speech Recognition API
  scope: per-request
  window: null
  limit: 512 MB audio / 2 GB video
  source: https://docs.mthreads.com/asr/recording-recognition/introduction
- id: asr-audio-duration
  api: Moore Threads AIBook Speech Recognition API
  scope: per-request
  window: null
  limit: 5 hours
  source: https://docs.mthreads.com/asr/recording-recognition/introduction
- id: asr-transcription-time
  api: Moore Threads AIBook Speech Recognition API
  scope: per-task
  window: null
  limit: 4 hours maximum transcription time
  source: https://docs.mthreads.com/asr/recording-recognition/introduction
- id: asr-result-retention
  api: Moore Threads AIBook Speech Recognition API
  scope: per-task
  window: 72 hours
  limit: Completed results retained 72 hours
  note: After expiry, query returns 3005 "cannot find the taskid".
  source: https://docs.mthreads.com/asr/recording-recognition/introduction
- id: asr-result-fetches
  api: Moore Threads AIBook Speech Recognition API
  scope: per-task
  window: lifetime of the task
  limit: 100 result retrievals
  source: https://docs.mthreads.com/asr/recording-recognition/introduction
- id: asr-upload-chunk
  api: Moore Threads AIBook Speech Recognition API
  scope: per-request
  window: null
  limit: 10 MB recommended per upload chunk
  source: https://docs.mthreads.com/asr/recording-recognition/introduction
- id: asr-realtime-pacing
  api: Moore Threads AIBook Speech Recognition API
  scope: per-connection
  window: 10 seconds
  limit: A gap of more than 10s between audio frames terminates the session
  note: Recommended pacing is 160 ms of audio every 160 ms. Sending faster than realtime returns 2004.
  source: https://docs.mthreads.com/asr/realtime-recognition/introduction
- id: vc-realtime-pacing
  api: Moore Threads AIBook Streaming Voice Conversion API
  scope: per-connection
  window: 10 seconds
  limit: Send exactly 1000 ms of audio per 1000 ms; a gap over 10s ends the session
  source: https://docs.mthreads.com/vc/stream-vc/introduction
undocumented:
- Requests per second / per minute on any AIBook surface.
- Concurrent connection ceiling (2003 exists but the number behind it is not published).
- Token or context limits on the KUAE Cloud Coding Plan endpoint.