What a call costs, and when it costs nothing

Tiers

Price follows the work

An endpoint's price reflects how many upstream requests it makes and how much assembly it does on top. The table below reads current prices and examples directly from the billing registry.

These tiers describe the catalogue; they are not a formula. The authoritative price of any single endpoint starts with the credit_cost field on its catalogue row — call GET /v1/utility/endpoints for all of them, or GET /v1/utility/endpoint?id=github/search for one. Both are free. Some operations quote or settle a variable charge (for example multi-source search, research and scheduled monitor runs); read their estimate and returned credits_used, not only the list price.

Never charged

What you do not pay for

Cache

A cached hit costs 0 credits

Public upstream data is cached briefly and shared across callers. When your call is answered from that cache, the net charge is 0 credits. A concurrent request that waited for another fetch may show a temporary reservation and matching refund. Request a new retrieval with fresh=1; when honoured, the uncached call follows that endpoint’s billing rules. Fresh bypasses have their own allowance; see caching. The envelope tells you: cached: true.

A temporary outage can also produce a marked older reference answer on amazon/categories, target/categories or sephora/brands. It costs 0 credits net, with stale: true and x-cache: STALE; the response includes the original retrieval time and age. The extra retention is at most one hour beyond the normal cache lifetime, with a total cap of 24 hours. Fresh requests rule this fallback out even when their fresh allowance is used, and max_age still applies. See the caching contract for its limits.

Paginated endpoints bill per page fetched, at the endpoint's price. Page-size defaults, maximums and any result-based or composite pricing depend on the endpoint. Check its parameters and pricing notes; there is no universal limit of 100.

Rate limits

Request limits grow with your plan

Every request must fit both its key and shared account limits. Starter key limits are shown below; Pro, Growth and Business increase request rate and concurrency. Buying extra credits does not change these limits. Individual endpoints may have lower source limits.

The limiter runs before any billing work, so a throttled request costs 0 credits. Batch items are admitted individually, including price previews, and consume request headroom. Compare every plan’s key and account limits. The failure mode is documented in errors.

Auditing

Check every credit

Each charge writes a ledger row referencing the request_id of the call that caused it, so any figure on your bill can be traced back to a single request.

$ curl "https://www.monocrawl.com/v1/credits/transactions?limit=5" \
    -H "x-api-key: mn_your_key_here"

"data": {
  "items": [
    { "id": "48211", "delta": -3, "balance_after": 94, "reason": "usage",
      "ref_type": "usage_event", "ref_id": "req_4f2b8c1d09ae37b562",
      "created_at": "2026-08-30T09:14:02.113Z" }
  ],
  "count": 1,
  "cursor": null
}