generated: '2026-08-13'
method: searched
source: >
  https://newsapi.ai/plans (server-prerendered, read 2026-08-13) for quota and plan limits,
  and Event Registry's own first-party client source for the runtime signalling —
  https://raw.githubusercontent.com/EventRegistry/newsapi-mcp/main/src/client.ts
  (parseTokenUsage reads the response headers),
  https://raw.githubusercontent.com/EventRegistry/newsapi-mcp/main/src/types.ts
  (classifyError maps status to category) and
  https://raw.githubusercontent.com/EventRegistry/newsapi-mcp/main/src/errors.ts
  (recovery guidance). The documentation portal at newsapi.ai/documentation is
  JavaScript-rendered and could not be read by machine.
name: Event Registry Rate Limits
description: Rate limiting policies for the Event Registry (NewsAPI.ai) news intelligence REST API.
url: https://newsapi.ai/documentation
rateLimits:
  - name: Concurrent Request Limit
    description: Maximum number of simultaneous API requests allowed per account at any given time.
    limit: 5
    window: concurrent
    scope: account
    enforcement: >
      If more than 5 simultaneous requests are made, the server returns HTTP 429 Too Many Requests.
      Excess requests are not queued — they are rejected immediately.
    recommendation: Make API requests sequentially rather than in parallel to avoid 429 errors.
  - name: Monthly Token Quota
    description: Each plan includes a monthly token allocation. API calls consume tokens based on the type of request and historical depth.
    scope: account
    enforcement: >
      Once the monthly token cap is reached, further API calls are rejected by default.
      Users can opt to allow overages (charged at $0.015 per token) by unchecking "Block API calls once the monthly cap is reached" in the dashboard.
    tokenConsumption:
      - action: Search recent articles (last 30 days)
        tokensPerCall: 1
      - action: Search historical articles
        tokensPerCall: 5 per year searched (e.g., searching 2017 costs 5 tokens; 2015-2017 costs 15 tokens)
      - action: Search recent events
        tokensPerCall: 5
      - action: Search historical events
        tokensPerCall: 20 per year searched
      - action: Per-document analytics
        tokensPerCall: 0.01 to 0.05
    overage:
      enabled: true (optional, opt-in)
      costPerToken: 0.015
      currency: USD
  - name: Free Plan Limit
    description: Free accounts are limited to 2,000 total search tokens with no monthly renewal.
    limit: 2000
    scope: account
    enforcement: Free plan tokens are a one-time allocation for testing purposes.
responseHeaders:
  - name: req-tokens
    description: Tokens consumed by THIS request. Present on successful responses.
    standard: false
    source: newsapi-mcp src/client.ts parseTokenUsage()
  - name: x-ratelimit-remaining
    description: Tokens remaining in the current quota period.
    standard: false
    source: newsapi-mcp src/client.ts parseTokenUsage()
missingHeaders:
  note: >
    Verified absent from the provider's own client, which reads only the two headers above.
    An agent gets a remaining-count and a per-request cost, but no limit, no reset time and
    no back-off hint.
  headers:
    - RateLimit / RateLimit-Policy (RFC 9331)
    - X-RateLimit-Limit
    - X-RateLimit-Reset
    - Retry-After
errorCodes:
  - code: 429
    category: rate_limit
    description: >
      TOKEN QUOTA exhausted — the monthly plan allocation (or the one-time free allocation)
      has been spent. This is NOT the concurrency signal.
    retryable: true
    resolution: >
      Do not retry immediately; retrying will not succeed. The provider's own guidance is
      that tokens refresh the next day, and monthly caps reset at the billing period. Either
      reduce consumption or enable overage at $0.015/token in the dashboard.
    source: newsapi-mcp src/types.ts classifyError + src/errors.ts
  - code: 503
    category: concurrent_limit
    description: >
      CONCURRENCY cap exceeded — more than 5 simultaneous requests from one account. Excess
      requests are rejected immediately, not queued.
    retryable: true
    resolution: Make requests sequentially, one after the other.
    source: >
      newsapi-mcp src/errors.ts, verbatim: "Too many simultaneous requests (HTTP 503).
      Event Registry allows max 5 concurrent requests. Make requests sequentially, one after
      the other." Recognised as a distinct category in newsapi-mcp 1.2.0 (2026-03-04).
  - code: 401
    category: auth_error
    description: Missing, malformed or invalid apiKey.
    retryable: false
    resolution: Check the key; obtain one at https://newsapi.ai/register.
correction:
  date: '2026-08-13'
  note: >
    A previous round of this artifact recorded 429 as the concurrency signal. That was wrong.
    Event Registry's own error classifier maps 429 to quota exhaustion and 503 to the
    5-concurrent-request cap. Corrected above.
authentication:
  method: API Key
  parameterName: apiKey
  placement: query parameter or request body
  acquisition: Register at https://newsapi.ai/register to obtain a free API key.
httpMethods:
  - GET
  - POST
notes:
  - POST is required when request content is too long for a GET URL.
  - All endpoints support both GET and POST unless otherwise specified.
  - The 5-request concurrency cap applies across all plan tiers.