Skip to content
Start free

Reference

Limits

Plan limits, and the sizes of records, requests, and settings.

Plans

PlanRecordsRecords in one indexDataSearches a monthIndexes
Free10,00010,00040 MB10,0001
Starter50,00050,000200 MB50,0003
Team150,000100,000600 MB750,00010
Growth500,000250,0002 GB3,000,00025
Scale1,000,000500,0004 GB10,000,00050

Each plan also has monthly ceilings far above what stores use, so a script sending events or rewriting records in a loop can't run up costs:

PlanEvents a monthRecord changes a monthAI allowance a month
Free100,000300,000$1
Starter1,000,0001,500,000$2
Team5,000,0004,500,000$4
Growth20,000,00015,000,000$8
Scale50,000,00030,000,000$15
  • Events are the views, clicks and orders sent with the events API, the SDKs and the WooCommerce and Shopify integrations. Past the ceiling, new events answer 202 with "paused": true and aren't stored until the next month; searches keep working.
  • Record changes are records written or deleted: 30 for each record the plan includes, every record rewritten daily. Records sent again unchanged don't count. Past the ceiling, writes answer 429 until the next month.
  • The AI allowance pays for reviews of searches that found nothing and for field suggestions on new indexes. Past it, they wait for the next month. The recommendations add-on and related results have budgets of their own.
  • Owners get an email at 80% of the events or record changes.

Paid plans include optimized images: 5 new resized copies a month for each record the plan includes.

PlanNew image copies a month
FreeNot included
Starter250,000
Team750,000
Growth2,500,000
Scale5,000,000
  • A copy is one image at one width, made the first time a page shows it. Copies already made load free and don't count, from any page or data center, and neither do placeholders.
  • Owners get an email at 80%. Past the month's copies, images without a copy yet load as their originals until the next month; copies already made keep loading.

Prices, spending limits, and what every plan includes are on the pricing page.

  • Data is the records' JSON in UTF-8, as you send it: 4 KB for each record a plan includes. Records of any size up to 100 KB fit, as long as the workspace's data does.

  • Searches are API and browser searches in a calendar month (UTC), including searches as you type. Searches answered from our cache, such as a popular query repeated, are free, and so are browser searches from verified crawlers such as Googlebot.

  • Free stops at its records, data and searches: writes that would go over, and searches after the month's are used, answer 402. Searches our cache can answer keep working.

  • Paid plans keep working past their records, data and searches. The extra is billed as overage with the next invoice, or with the renewal on yearly plans:

    • extra searches are $0.20 per 1,000 in a month;
    • extra records are $1 per 10,000 a month, counted by the most held on a day of each month, on yearly plans too. Data past what the plan includes counts each 4 KB as a record, when that comes to more than the records themselves;
    • each never costs more in a month than a larger plan would, with that plan's own extra. Starter with a million searches costs what Team would.

    Extra searches have a spending limit: by default the difference to the next plan's price, which covers up to that plan's searches (Starter's $30 covers 750,000). Owners change it under Settings, from $0 to no limit. Once a month's searches reach what it covers, searches our cache can't answer get 402 until the next month or a higher limit; verified crawlers' searches still work.

    Records in one index are a hard limit on every plan. An index's subindexes share it: the index and its subindexes together.

  • Buying a plan: owners choose and change plans under Settings in the dashboard, billed monthly or yearly (two months free) through our reseller, Polar. The billing portal there has invoices and cancellation.

    • A larger plan, or yearly billing, starts at once.
    • A smaller plan starts at renewal, once the workspace fits its indexes and records per index.
    • Each person can have one workspace on Free; other workspaces start with a paid plan.
  • Members are the same on every plan: a workspace can have up to 50, counting pending invitations. Each person can send up to 20 new invitations a day.

The dashboard shows the workspace's usage under Settings.

Records

Limit
objectID1 to 256 characters, other than . and ..
Record size100 KB of JSON
Fields in a record256
Field path depth, for fields in settings8 levels, such as a.b.c
Text in a searchable field1 KB in the two short fields, 16 KB in the others; see Searchable fields

Requests

Limit
Records in an upsert1 to 1,000
Records in a replaceAll100,000
Records in a staged replacement batch1 to 1,000
Query length200 characters
Filters20 fields, and 100 values in an in filter
Facets32 a request
facetStats8 number fields a request
Hits a request100 from your server, 50 from browsers; 0 with clusters
Offset5,000
attributesToRetrieve64 fields
Highlight tags32 characters each
near radius1 m to 20,000 km
withinOne box a request
tileOne tile a request, zoom 0 to 22; each one the cache doesn't answer counts as one search
clusters1, 2, 4, 8 or 16 cells a side; stats on 3 number fields
spread1, 2, 4, 8 or 16 cells a side; 10 records a cell, up to the request's hits
Browser searches240 a minute per visitor, 6,000 a minute per index, by default; see Bots and rate limits
API-key searches3,000 a minute per key by default; 600 or 12,000 on the API keys page
Searches running at onceA share of capacity for each workspace, larger on larger plans; one over it waits a moment, then gets 503
Search time10 seconds a query, then 503
Recommendations20 products a list; not counted as searches; browser requests share browser searches' rate limits
Lists3 per index on Free, 25 on paid plans; 50 products or 1,000 values a list; 10 lists a request; 300 requests a minute per visitor from browsers; not counted as searches
Events100 a request from browsers and 200 from servers; 600 requests a minute per visitor or per key; not counted as searches
Product feedsOne per index; up to 1 GB and 1,000,000 items; read at most hourly; the last 20 runs kept
UpdatesHourly totals kept 90 days; the latest 500 operations and 200 refused writes; /writes/log 200 a request
Subindexes5 per index, not counted among the plan's indexes; 5 a search, which counts once with them
Multi-index search10 searches a request, counted as one search for each workspace
Query suggestions10 a request; 600 requests a minute per visitor from browsers; an API key's count toward its searches a minute; not counted as searches
Related resultsPaid plans; indexes of up to 200,000 records; up to 50 a search that found nothing, a page at a time as limit asks; embedding capped at 3% of the plan's price a month

Settings

Limit
Searchable fields1 to 5, with weights from 0.1 to 100
Filter, facet, and sort fields128
Sorts8, besides relevance
Custom ranking3 fields
Location fields1 an index, with one location a record
Synonyms1,000 groups of 2 to 20 terms, each up to four words and 64 characters
Allowed origins20 an index