generated: '2026-08-13'
method: searched
source: https://docs-dev.markable.ai/#rate-limits
limit_count: 1
rate_limits:
- scope: per-client
  window: 1m
  limit: 25
  unit: requests
  burst: null
  applies_to: search results per minute, per client
  quote: 'For have fair usage of our APIs, we have enforced rate limits to the number of
    search results one can perform per minute. For each client, the current limit is 25
    requests/minute.'
  increase: 'documented as negotiable — "Depending on your use case, you can get in touch to
    have your limits increased."'
  source: https://docs-dev.markable.ai/#rate-limits
response_signaling:
  headers: []
  headers_note: >-
    Markable documents no rate-limit response headers (no X-RateLimit-*, no RateLimit-*, no
    Retry-After). An agent cannot read remaining quota from a response; it can only observe the
    exhaustion status code.
  exhaustion_status: 429
  exhaustion_body: 'JSON:API error object array — {"errors":[{"title":"Too Many Requests",
    "status":429,...}]}'
  exhaustion_source: https://docs-dev.markable.ai/#http-error-codes
payload_limits:
- surface: image search input
  constraints:
  - 'accepted formats: .jpg, .jpeg, .png'
  - 'minimum image dimension: 100 px'
  - 'maximum image dimension: 6000 px'
  - 'maximum image size: 10485760 bytes (10 MiB)'
  - 'min/max image aspect ratio: 0.4 to 2.5'
  source: https://docs-dev.markable.ai/#search-constraints
notes:
- >-
  Published limits could not be observed on a live response — the documented API hosts are not
  currently serving (auth.markable.ai and video.markable.ai do not resolve; catalog.markable.ai
  answers 404 from an unrouted Kubernetes ingress with a default self-signed certificate,
  probed 2026-08-13). Values here are the vendor's published numbers, not observed ones.
- >-
  The consumer plans at https://www.markable.ai/pricing carry product quotas (Auto DMs, AI
  collages, AI captions) rather than API rate limits; those are recorded in
  plans/markable-plans-pricing.yml.