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: MiniMax
providerId: minimax-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
- Multimodal
- Rate Limiting
- Quotas
description: >-
  Per-modality rate limits for the MiniMax API. Text APIs share a single 500 RPM / 20M TPM
  pool; speech, video, image, and music APIs each have their own caps.
notes: Higher limits available on request via api@minimax.io.
sources:
- https://platform.minimax.io/docs/guides/rate-limits
responseCodes:
  throttled: 429
limits:
- name: Text APIs (all models)
  scope: account
  metric: requests-per-minute
  limit: 500
  notes: Combined with 20,000,000 TPM (input + output).
- name: Text APIs TPM
  scope: account
  metric: tokens-per-minute
  limit: 20000000
- name: Speech (T2A)
  scope: account
  metric: requests-per-minute
  limit: 60
  notes: 20,000 TPM.
- name: Voice Cloning
  scope: account
  metric: requests-per-minute
  limit: 60
- name: Voice Design
  scope: account
  metric: requests-per-minute
  limit: 20
- name: Video Generation
  scope: account
  metric: requests-per-minute
  limit: 5
- name: Image Generation
  scope: account
  metric: requests-per-minute
  limit: 10
  notes: 60 TPM.
- name: Music Generation
  scope: account
  metric: requests-per-minute
  limit: 120
  notes: Max 20 concurrent connections.
policies:
- name: Backoff Strategy
  description: Exponential backoff with jitter; honor Retry-After.
- name: Limit Increase
  description: Contact api@minimax.io for higher caps.
maintainers:
- FN: Kin Lane
  email: kin@apievangelist.com