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: Synthflow
providerId: synthflow
generated: '2026-08-13'
method: searched
source: https://docs.synthflow.ai/concurrency-calling-limits
created: '2026-05-22'
modified: '2026-08-13'
tags:
  - Rate Limiting
  - Concurrency
  - CAPS
  - Telephony
description: >-
  Synthflow documents two throttles, and both are call-shaped rather than
  request-shaped: Concurrent Active Calls (how many calls can be live at once) and CAPS,
  Call Attempts Per Second (how fast new calls can be started through the Make a Phone
  Call endpoint). Neither has a published number — both are scoped per account or shown in
  workspace/account terms. There is no documented general HTTP request rate limit for the
  other 99 operations, and no rate-limit response headers are published.

limit_count: 2
numeric_limits_published: false
headers_published: false

limits:
  - name: Concurrent Active Calls
    scope: workspace
    metric: concurrent_calls
    limit: null
    limit_disclosure: 'Enterprise or account-managed deployments: scoped with Synthflow. Existing legacy workspaces: shown in workspace or account terms.'
    timeFrame: instantaneous
    applies:
      - https://api.synthflow.ai/v2
      - https://api.us.synthflow.ai/v2
      - https://api.eu.synthflow.ai/v2
    description: >-
      Caps how many live calls can run simultaneously. When one call ends another can
      begin. Legacy Pro/Growth/Agency plans carried fixed concurrency limits and remain
      supported for existing customers.
  - name: CAPS (Call Attempts Per Second)
    scope: workspace
    metric: call_attempts_per_second
    limit: null
    limit_disclosure: 'Enterprise or account-managed deployments: scoped with Synthflow. Existing legacy workspaces: shown in workspace or account terms.'
    timeFrame: second
    applies_to_operations: [voice-call]
    endpoint: POST /v2/calls
    description: >-
      Caps how frequently new outbound calls can be initiated, independent of how many are
      active. Provider example: at CAPS 2, sending 10 call requests in one second exceeds
      the limit and some are rejected or queued.

general_api_rate_limit:
  published: false
  note: >-
    No documented per-key or per-endpoint request rate limit covering the non-calling
    operations (agents, actions, knowledge bases, contacts, memory stores, simulations,
    webhook logs). An integrator cannot size a bulk read against a published budget.

responseCodes:
  throttled: 429
  note: >-
    "Depending on your account setup, excess calls can return a 429 HTTP error or be queued
    briefly." The behaviour is account-dependent, so 429 is not a guaranteed signal — a
    caller may instead observe added latency. 429 is declared on ZERO of the 100 operations
    in openapi/_original/synthflow-openapi.json.

headers:
  limit: null
  remaining: null
  reset: null
  retryAfter: null
  policy: null
  note: >-
    No RateLimit-*, X-RateLimit-* or Retry-After header contract is published anywhere in
    the documentation, and no rate-limit header is declared in the OpenAPI. This is the
    material gap: an agent gets no runtime budget signal and can only discover the ceiling
    by hitting it.

recoveryStrategies:
  - name: Client-side pacing
    description: >-
      Because no Retry-After is published, pace outbound call creation below the
      contracted CAPS rather than reacting to a 429.
  - name: Exponential backoff
    description: 'Back off on 429 and retry. Note there is no idempotency key, so a retry of voice-call risks placing a duplicate call — see conventions/synthflow-conventions.yml.'

custom_limits:
  available: true
  process: 'Enterprise and account-managed customers can request custom concurrency or CAPS limits.'
  contact: https://synthflow.ai/talk-to-sales

notes:
  - >-
    This file replaces a scaffold that asserted 10 req/min free and 120 req/min pro tiers
    with X-RateLimit-* headers. None of those values or headers are published by Synthflow.
  - 'Related telephony guidance: https://docs.synthflow.ai/telephony-quality-and-performance'

x-evidence:
  fetched: '2026-08-13'
  probes:
    - {url: 'https://docs.synthflow.ai/concurrency-calling-limits.md', status: 200}
    - {url: 'https://docs.synthflow.ai/cuncurrency-calling-limits.md', status: 200, detail: 'legacy misspelled slug, now a not-found page that redirects readers to the correct page'}