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: Datadog APM
providerId: datadog-apm
generated: '2026-09-05'
method: searched
source: >-
  https://docs.datadoghq.com/api/latest/rate-limits/ and
  https://docs.datadoghq.com/bits_ai/mcp_server/ (MCP fair-use limits)
created: '2026-05-04'
modified: '2026-09-05'
supersedes: >-
  A 2026-05-04 bulk-sweep scaffold that invented per-tier numeric limits (10/min free, 100/min pro,
  1000/min enterprise, monthly request quotas). Datadog publishes none of those. Every value below is
  read from Datadog's own documentation, and where Datadog declines to publish a number that is
  recorded as an explicit gap rather than filled in.
tags:
- APM
- Rate Limiting
- Throttling
description: >-
  Datadog rate-limits per endpoint bucket, not per plan tier, and it declines to publish the numeric
  limits — the documentation says explicitly that the list of limits is not comprehensive and that the
  authoritative value for any endpoint arrives in the response headers. For an agent this is actually
  workable: the runtime signal is complete and self-describing even though the static number is not.
headers:
  limit: X-RateLimit-Limit
  period: X-RateLimit-Period
  remaining: X-RateLimit-Remaining
  reset: X-RateLimit-Reset
  name: X-RateLimit-Name
header_semantics:
- header: X-RateLimit-Limit
  meaning: Number of requests allowed in a time period.
- header: X-RateLimit-Period
  meaning: Length of time in seconds for resets (calendar aligned).
- header: X-RateLimit-Remaining
  meaning: Number of allowed requests left in the current time period.
- header: X-RateLimit-Reset
  meaning: Time in seconds until next reset.
- header: X-RateLimit-Name
  meaning: >-
    Name of the rate limit — the bucket identifier to quote when asking Datadog to raise it. No other
    major observability API publishes this, and it is the field that makes an increase request
    actionable.
retry_after_header: false
responseCodes:
  throttled: 429
limit_count: 3
limits:
- name: Event submission
  scope: organization
  metric: events_per_minute
  limit: 250000
  timeFrame: minute
  applies:
  - Datadog Events API
  source: https://docs.datadoghq.com/api/latest/rate-limits/
- name: Datadog MCP Server tool call burst
  scope: mcp-session
  metric: requests_per_10_seconds
  limit: 50
  timeFrame: 10 seconds
  applies:
  - https://mcp.datadoghq.com/api/unstable/mcp-server/mcp
  source: https://docs.datadoghq.com/bits_ai/mcp_server/
  note: Datadog labels this fair use and subject to change.
- name: Datadog MCP Server monthly tool calls
  scope: organization
  metric: tool_calls_per_month
  limit: 100000
  timeFrame: month
  applies:
  - https://mcp.datadoghq.com/api/unstable/mcp-server/mcp
  source: https://docs.datadoghq.com/bits_ai/mcp_server/
not_rate_limited:
- surface: Metric / data point submission
  note: >-
    Not rate limited. Constrained instead by the custom metrics volume in the commercial agreement —
    a billing limit, not a throttle.
- surface: Log submission
  note: Not rate limited.
undocumented:
  per_endpoint_limits: >-
    The numeric limit for the APM endpoints in openapi/ (spans search, span analytics, SLO reads and
    writes, retention filters) is NOT published. Datadog states limits vary per endpoint and are
    returned in the headers. Recorded as an honest gap — an agent cannot pre-compute a safe call rate
    for these operations from any published document.
  declared_in_spec: >-
    Confirming the same thing from the other side: 429 is declared on all 34 operations in openapi/,
    and none of them declares a limit value.
policies:
- name: Backoff
  description: >-
    Datadog's documented remediation is to wait the X-RateLimit-Period, or to call at a frequency
    slightly longer than X-RateLimit-Limit / X-RateLimit-Period. There is no Retry-After header on
    this API, so X-RateLimit-Reset is the value to sleep on.
- name: Increase on request
  description: >-
    Limits can be raised from the defaults by contacting Datadog support; quote the X-RateLimit-Name
    value of the bucket being hit.
observability:
  self_service: true
  description: >-
    Datadog exposes API rate limit consumption as usage metrics (allowed and blocked requests, by
    endpoint dimension), so an organization can audit its own limit headroom in Datadog itself. It
    excludes metric, log and event submission endpoints.
  dashboard: https://app.datadoghq.com/dashboard/lists?q=Datadog%20api%20rate%20limit%20usage
maintainers:
- FN: Kin Lane
  email: kin@apievangelist.com