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: Contextual AI
providerId: contextual-ai
created: '2026-06-20'
modified: '2026-06-20'
reconciled: false
tags:
- AI
- RAG
- LLM
- Grounded Language Model
- Enterprise
- Rate Limiting
- Quotas
- Throttling
description: >-
  The Contextual AI platform enforces per-workspace rate limits and request
  constraints across its APIs. Documented hard input constraints include a
  32,000-token total limit on Generate requests, a 7,000-token total limit on
  LMUnit requests, and Parse file limits of 300 MB and 2,000 pages per file.
  Per-endpoint request-per-minute and token-per-minute throttles vary by tier
  and are not publicly reconciled in this artifact; enterprise agreements raise
  limits. Throttled requests return HTTP 429.
notes: >-
  Capture per-endpoint RPM/TPM and concurrency limits from the Contextual AI
  console or enterprise agreement during reconciliation. Documented constraints
  here are the request-size limits stated in the API reference (Generate 32k
  tokens, LMUnit 7k tokens, Parse 300 MB / 2000 pages, parse results retained
  30 days).
sources:
- https://docs.contextual.ai/api-reference/generate/generate
- https://docs.contextual.ai/api-reference/lmunit/lmunit
- https://docs.contextual.ai/api-reference/parse/parse-file
responseCodes:
  throttled: 429
limits:
- name: Generate Request Tokens
  scope: request
  metric: tokens
  limit: 32000
  notes: Total tokens (messages + knowledge + output) per Generate request.
- name: LMUnit Request Tokens
  scope: request
  metric: tokens
  limit: 7000
  notes: Total input tokens per LMUnit evaluation request.
- name: Parse File Size
  scope: request
  metric: bytes
  limit: 314572800
  notes: Maximum 300 MB per file submitted to Parse.
- name: Parse File Pages
  scope: request
  metric: pages
  limit: 2000
  notes: Maximum 2,000 pages per file submitted to Parse.
- name: Requests Per Minute (RPM)
  scope: workspace
  metric: requests
  limit: see provider documentation
  notes: Per-endpoint RPM varies by tier; not publicly reconciled.
- name: Tokens Per Minute (TPM)
  scope: workspace
  metric: tokens
  limit: see provider documentation
  notes: Per-endpoint TPM varies by tier; not publicly reconciled.
policies:
- name: Job Result Retention
  description: Parse job status and results are retained for 30 days; older requests return 404.
- name: Tiered Limits
  description: Limits raise with paid usage and via Enterprise agreements.
- name: Backoff Strategy
  description: Clients should implement exponential backoff with jitter and honor Retry-After on 429.
maintainers:
- FN: Kin Lane
  email: kin@apievangelist.com