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: Moonshot AI
providerId: moonshot-ai
created: '2026-05-08'
# Provenance stamped 2026-08-11: this artifact was written by the API Evangelist
# bulk sweep dated 2026-05-08, not harvested from the provider. See roadmap#35.
method: generated
modified: '2026-05-08'
reconciled: true
tags:
- AI
- LLM
- Inference
- Long Context
- Kimi
- Rate Limiting
- Quotas
- Throttling
description: >-
  Tiered rate limits for Moonshot AI / Kimi platform. Tier0 starts at $1 cumulative recharge
  with very low limits; tiers progress to Tier5 ($3,000 cumulative) with up to 1,000 concurrency,
  10K RPM, 5M TPM, and unlimited TPD.
notes: >-
  Vouchers do not count toward cumulative recharge. Cluster capacity events may temporarily
  reduce effective ceilings. Contact api-service@moonshot.ai for higher limits.
sources:
- https://platform.kimi.ai/docs/pricing/limits
responseCodes:
  throttled: 429
limits:
- name: Tier0 (>= $1 recharge)
  scope: account
  metric: requests-per-minute
  limit: 3
  notes: Concurrency 1, TPM 500K, TPD 1.5M.
- name: Tier1 (>= $10 recharge)
  scope: account
  metric: requests-per-minute
  limit: 200
  notes: Concurrency 50, TPM 2M, TPD unlimited.
- name: Tier5 (>= $3000 recharge)
  scope: account
  metric: requests-per-minute
  limit: 10000
  notes: Concurrency 1000, TPM 5M, TPD unlimited.
- name: Concurrency Tier0
  scope: account
  metric: concurrent-requests
  limit: 1
- name: Concurrency Tier5
  scope: account
  metric: concurrent-requests
  limit: 1000
policies:
- name: Backoff Strategy
  description: Clients should implement exponential backoff with jitter and honor Retry-After.
- name: Tier Upgrade
  description: Cumulative recharge advances tier eligibility automatically; vouchers excluded.
maintainers:
- FN: Kin Lane
  email: kin@apievangelist.com