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: Mixedbread
providerId: mixedbread-ai
created: '2026-05-25'
modified: '2026-05-25'
reconciled: true
tags:
  - AI
  - Rate Limiting
  - Quotas
description: Reconciled rate limits for the Mixedbread platform. Limits are plan-tier driven and split between
  query and ingestion lanes; Enterprise tier is custom.
sources:
  - https://www.mixedbread.com/pricing
  - https://www.mixedbread.com/api-reference
algorithm: token-bucket
responseCodes:
  throttled: 429
  quotaExceeded: 429
headers:
  retryAfter: retry-after
limits:
  - tier: Starter
    plan: Free
    lane: combined
    rpm: 100
    notes: Combined request budget across all endpoints
  - tier: Scale
    plan: Scale ($20/mo)
    lane: queries
    rpm: 1200
    notes: Query lane — embeddings, reranking, store search, question-answering
  - tier: Scale
    plan: Scale ($20/mo)
    lane: ingestion
    rpm: 360
    notes: Ingestion lane — file uploads, parsing, extraction, store writes
  - tier: Enterprise
    plan: Enterprise
    lane: custom
    rpm: -1
    notes: Custom rate limits and SLA defined in contract
quotas:
  - tier: Starter
    resource: stores
    limit: 10
  - tier: Scale
    resource: stores
    limit: 10000
  - tier: Enterprise
    resource: stores
    limit: -1
notes:
  - 429 responses include a retry-after hint.
  - Burst behavior is governed by the platform; enterprise customers can negotiate per-region capacity.
  - Hugging Face Inference Providers usage (Mixedbread is a partner provider) is rate-limited by the Hugging Face
    platform, not by api.mixedbread.com.