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: Ximilar
providerId: ximilar
created: '2026-05-04'
modified: '2026-08-28'
generated: '2026-08-28'
method: searched
source: https://www.ximilar.com/pricing/ + https://docs.ximilar.com/quickstart + https://docs.ximilar.com/async-api
docs: https://www.ximilar.com/pricing/
tags:
  - Rate Limiting
  - Quotas
  - Credits
description: >-
  Published limits for the Ximilar API. This file REPLACES a 2026-05-04 scaffold that
  asserted per-minute request ceilings and X-RateLimit-* response headers Ximilar has never
  documented. The honest picture: Ximilar throttles by CREDIT QUOTA and by resource count,
  not by requests per second, and it publishes NO rate-limit response headers and NO 429
  contract. On the synchronous surface a caller has no runtime backoff signal at all; on the
  asynchronous surface throttling surfaces as a request STATE (RETRY, API_LIMIT_EXCEEDED)
  rather than an HTTP status.
limit_count: 9
headers:
  published: false
  limit: null
  remaining: null
  reset: null
  retryAfter: null
  policy: null
  note: >-
    No X-RateLimit-*, RateLimit-* or Retry-After header is documented anywhere in the docs
    corpus (checked against docs.ximilar.com/llms-full.txt, 614KB, on 2026-08-28).
responseCodes:
  throttled: null
  quotaExceeded: null
  note: >-
    No HTTP status for throttling or quota exhaustion is documented. The only published
    exhaustion signal is on the async request API, where the request status becomes
    API_LIMIT_EXCEEDED ("Rate limit hit or insufficient credits") or RETRY ("a transient
    failure occurred, e.g. rate limit or GPU OOM; the request will be retried automatically").
limits:
  - tier: free
    name: Free monthly credit allowance
    scope: account
    metric: api_credits_per_month
    limit: 1000
    timeFrame: month
    source: https://www.ximilar.com/pricing/
  - tier: business
    name: Business monthly credit allowance
    scope: account
    metric: api_credits_per_month
    limit: 500000
    limit_range: [100000, 300000, 500000]
    timeFrame: month
    source: https://www.ximilar.com/pricing/
    note: Sold as three fixed steps — Business 100K / 300K / 500K.
  - tier: professional
    name: Professional monthly credit allowance
    scope: account
    metric: api_credits_per_month
    limit: 40000000
    limit_range: [1000000, 2000000, 3000000, 4000000, 5000000, 6000000, 7000000, 10000000, 20000000, 30000000, 40000000]
    timeFrame: month
    source: https://www.ximilar.com/pricing/
    note: Eleven fixed steps from 1M to 40M credits per month.
  - tier: all
    name: Batch size per request
    scope: request
    metric: records_per_request
    limit: 10
    timeFrame: request
    source: https://docs.ximilar.com/quickstart
    note: >-
      "This allows you to send up to 10 images (records) in a single request via the
      `records` field." Each record is billed individually.
  - tier: free
    name: Visual search collection size
    scope: collection
    metric: records_per_collection
    limit: 1000
    source: https://www.ximilar.com/pricing/
  - tier: business
    name: Visual search collection size
    scope: collection
    metric: records_per_collection
    limit: 500000
    limit_range: [100000, 300000, 500000]
    source: https://www.ximilar.com/pricing/
  - tier: professional
    name: Visual search collection size
    scope: collection
    metric: records_per_collection
    limit: -1
    source: https://www.ximilar.com/pricing/
    note: Published as "unlimited".
  - tier: all
    name: Number of collections
    scope: account
    metric: collections
    limit: 10
    limit_by_tier: {free: 1, business_100k: 3, business_300k: 5, business_500k: 8, professional: 10}
    source: https://www.ximilar.com/pricing/
  - tier: all
    name: VLM request history page size
    scope: request
    metric: results_per_page
    limit: 100
    default: 20
    source: https://docs.ximilar.com/platform/vlm
    note: 'GET /vlm/v2/request/history/ — page_size default 20, max 100.'
policies:
  - name: Credit exhaustion
    description: >-
      Once the monthly credit allowance is spent, calls that consume credits fail. Additional
      credit packs can be purchased on Business and Professional plans. Only calls that send
      or upload an image for AI analysis consume credits.
    source: https://docs.ximilar.com/quickstart
  - name: Automatic retry on the async surface
    description: >-
      Asynchronous requests that hit a transient failure — explicitly including a rate limit
      — are moved to the RETRY state and retried by Ximilar automatically, with no client
      action required.
    source: https://docs.ximilar.com/async-api
  - name: Storage capacity
    description: >-
      Inserts into a collection that exceed the plan's collection size are refused with
      status 412 "hard capacity exceeded".
    source: https://docs.ximilar.com/visual-search/overview
gaps:
  - No per-second, per-minute or per-hour request ceiling is published for any tier.
  - No rate-limit response headers and no 429 contract are documented.
  - >-
    An agent calling the synchronous services has no published runtime signal to back off on;
    the only throttling feedback Ximilar documents lives on the asynchronous request API.
maintainers:
  - FN: Kin Lane
    email: kin@apievangelist.com