generated: '2026-08-16'
method: searched
source: https://vehicles.dev/docs#pricing
limit_count: 3
algorithm: token bucket
scope: per account
scope_note: >-
  Rate limits and credit balances are enforced per account, not per key, so splitting traffic across
  several API keys does not raise the throughput ceiling. The bucket refills at the plan's
  requests-per-second rate.
limits:
  - scope: per-account
    plan: Starter
    limit: 5
    window: 1 second
    unit: requests
    burst: null
    burst_note: Token bucket; the docs publish the refill rate but not a separate burst capacity.
  - scope: per-account
    plan: Pro
    limit: 10
    window: 1 second
    unit: requests
    burst: null
  - scope: per-account
    plan: Scale
    limit: 50
    window: 1 second
    unit: requests
    burst: null
exhaustion:
  status: 429
  code: rate_limit_exceeded
  retryable: true
  body: application/problem+json (RFC 9457)
  billing: >-
    The billing reservation is released on a 429, so the rejected call is free.
response_headers:
  standard:
    - header: retry-after
      present_on: 429
      units: whole seconds
      description: How long to wait before retrying. The only rate-limit signal the API emits.
    - header: x-request-id
      present_on: every response, success or failure
      description: Server-generated UUID; identical to request_id in a problem body. Client-supplied values are ignored and never echoed.
    - header: cache-control
      present_on: varies
  ratelimit_headers_published: false
  ratelimit_headers_note: >-
    The docs state explicitly: "There are no x-ratelimit-* headers." No RateLimit-Limit /
    RateLimit-Remaining / RateLimit-Reset (RFC 9331 draft) either. The complete set of non-standard
    headers this API sets is x-request-id, retry-after and cache-control, so an agent cannot read
    remaining quota — it can only react to a 429.
quota_signals:
  included_calls_header: false
  credit_balance_header: false
  note: >-
    No machine-readable entitlement or credit balance. 402 insufficient_credits is the only
    programmatic exhaustion signal for metered spend; usage is visible only in the dashboard.
retry_guidance:
  source: https://vehicles.dev/docs#errors-retries
  guidance: >-
    Branch on the `retryable` boolean in the problem document rather than the status code. Every 503
    and the 429 set it true; other 4xx are false. Retry with exponential backoff and jitter. All ten
    data endpoints are GETs and mutate nothing, so a retried read is always safe.
related:
  - conventions/vehicles-dev-api-conventions.yml
  - errors/vehicles-dev-api-problem-types.yml
  - plans/vehicles-dev-api-plans-pricing.yml