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: Cognition Labs
providerId: cognition-labs
created: '2026-07-02'
modified: '2026-07-02'
reconciled: false
tags:
- AI
- AI Agent
- Devin
- Rate Limiting
- Quotas
description: >-
  Cognition does not publish fixed numeric per-minute or per-day request-rate
  limits for the Devin API (v1, v2, or v3). Instead, usage is governed by
  Agent Compute Unit (ACU) consumption against a plan's quota or an
  Enterprise order-form allowance: Devin explicitly documents "no concurrent
  session limits," so the practical ceiling on how much work can run in
  parallel is a customer's own ACU budget and Enterprise-configured
  per-organization consumption caps rather than a request-throttling gate.
  Structured output schemas are capped at 64KB and session tags at 50 entries,
  which are the only numeric ceilings found in the documented request payloads.
notes: >-
  No documented endpoint returns rate-limit headers (e.g. X-RateLimit-*) and
  no 429 behavior is described in the public API reference or the Common
  Issues troubleshooting page as of the review date; verify directly with
  Cognition/Devin support before assuming request-level throttling exists.
sources:
- https://docs.devin.ai/api-reference/overview
- https://docs.devin.ai/api-reference/authentication
- https://docs.devin.ai/admin/billing/usage
- https://docs.devin.ai/admin/billing/enterprise
- https://docs.devin.ai/admin/common-issues
responseCodes:
  throttled: not documented
limits:
- name: Concurrent Sessions
  scope: account
  metric: sessions
  limit: none documented
  notes: Devin's own docs state there are no concurrent session limits; work can be split across multiple simultaneous sessions.
- name: Plan ACU Quota
  scope: account
  metric: acu
  limit: per plan (daily and/or weekly, see Plans)
  notes: Pro and Teams full seats have daily and weekly allowances; Max has a weekly-only allowance with no daily cap.
- name: Enterprise Organization Consumption Cap
  scope: organization
  metric: acu
  limit: admin-configured
  notes: Enterprise admins can set a per-organization ACU cap; when reached, all Devin activity in that organization stops until an admin intervenes.
- name: Session Tags
  scope: request
  metric: tags
  limit: 50
  notes: Maximum of 50 tags accepted on session create and on the update-tags endpoint.
- name: Structured Output Schema Size
  scope: request
  metric: bytes
  limit: 65536
  notes: JSON Schema (Draft 7) passed as structured_output_schema on session create is capped at 64KB.
- name: API Request Rate
  scope: account
  metric: requests
  limit: not published
  notes: No numeric per-minute/per-day request rate limit or 429 policy is documented for v1, v2, or v3 endpoints.
policies:
- name: ACU-Gated Rather Than Request-Gated
  description: The primary throttle on Devin usage is ACU quota/consumption-cap exhaustion, not HTTP request-rate limiting.
- name: Sleep on Inactivity
  description: A session enters sleep mode after roughly 0.1 ACU of inactivity and consumes zero units while asleep; sending a message wakes it.
- name: Validation Before Throttling
  description: Documented failure modes for these endpoints are 422 validation errors on malformed payloads, not 429 throttling responses.
maintainers:
- FN: Kin Lane
  email: kin@apievangelist.com