generated: '2026-08-13'
method: searched
source: https://developers.teads.com/
description: >-
  Rate limiting posture for the Teads developer surfaces. An exhaustive read of
  the Teads developer portal (all 93 pages in its sitemap were fetched and
  searched on 2026-08-13) found NO published rate limit, quota, throttling
  policy, 429 status documentation, or rate-limit response header. This is an
  honest zero, not an unchecked field.
limit_count: 0
limits: []
response_headers:
  documented: false
  observed: []
  note: >-
    No X-RateLimit-*, RateLimit-* or Retry-After header is documented for either
    the In-Chat Recommendations API or the Conversions API, and neither endpoint
    can be exercised anonymously to observe one.
exhaustion_status_code: undocumented
evidence:
- url: https://developers.teads.com/sitemap.xml
  status: 200
  note: 93 documentation URLs enumerated and fetched.
- url: https://developers.teads.com/docs/Chatbot-AI-SDK/Getting-Started/integration-guide/
  status: 200
  note: >-
    Full In-Chat API V2.0 technical specification — endpoint, headers, query
    parameters, request body, response schema and tracking lifecycle — with no
    rate limit, quota or throttling section.
- url: https://developers.teads.com/docs/Chatbot-AI-SDK/Features/settings/
  status: 200
  note: >-
    Configuration reference documents request timeouts as a client-side setting
    but publishes no server-side limit.
- url: https://ca.teads.tv/openapi.json
  status: 404
  note: No machine-readable contract to derive limits from.
practical_controls:
  note: >-
    The only volume-shaping controls Teads documents are commercial and
    placement-level rather than API-level — each widgetJSId fixes the number of
    recommendations returned, and a bid floor (bf / floorPrice) constrains which
    demand is eligible. Neither is a request-rate limit.
  fields:
  - widgetJSId
  - bf
  - floorPrice
recommendation: >-
  Teads should publish a rate limit for the In-Chat Recommendations API and emit
  RateLimit-* / Retry-After headers. Conversational integrations call this
  endpoint on every user turn, so an undocumented limit is a production risk for
  every partner.