Skip to content
Easysearch Docs

Limits

Number of pages indexed

The amount of pages that can be made available for your users to search through at once.

  • This is a hard limit; if you hit this limit, no more pages will be crawled until you remove sources to place yourself under the limit.
  • If you downgrade your plan and your new plan’s pages limit is lower than your current amount of indexed pages, pages will be deleted to move you under the limit.
  • If you are at the limit, existing pages can still be refreshed using the automatic refresh feature.
  • If you downgrade your plan, the new plan’s limit is immediately enforced. If necessary, pages will be deleted to ensure you remain under this limit. If your new limit is N pages, your pages will be sorted by crawl date (oldest first), and then the first N pages will be kept.

The limit depends on your current plan:

  • Free: 500
  • Small: 10,000
  • Medium: 25,000
  • Large: 50,000

Crawls per month

The amount of times Easysearch loads a page from your site in an attempt to index it.

Things to know:

  • If Easysearch follows a link that redirects to a page that was already crawled, a crawl will still be counted. However, it will remember that redirect, so future attempts will not incur an extra crawl.
  • If the page does not contain text content, a crawl will still be counted.
  • A crawl is counted when an existing page is refreshed.
  • This limit applies on a rolling basis. When making a crawl attempt, if you have hit your plan’s limit over the past 30 days, the crawl attempt will be delayed until after you are under your plan’s limit. Crawls will resume immediately after you return to below your plan’s limit.

This limit depends on your current plan:

  • Free: 5,000/mo (≈ 7/hour)
  • Small: 20,000/mo (≈ 28/hour)
  • Medium: 100,000/mo (≈ 2/minute)
  • Large: 1,000,000/mo (≈ 23/minute)

Search queries per month

The amount of times your users get search results every month.

Things to know:

  • Keyword search and Search by Meaning collectively count as 1 query when combined into a single request.
  • If you enable Search As You Type, every time results refresh will count as 1 query. Increasing the delay will decrease the amount of queries received, but it will make your user interface feel less responsive. Adjust this accordingly, and disable it if your usage gets too high.
  • This limit applies on a rolling basis. When making a search query, if you have hit your plan’s limit over the past 30 days, the request will be denied. When you return to below your plan’s limit, user search queries may resume immediately.

This limit depends on your current plan:

  • Free: 50,000/mo (≈ 1/minute)
  • Small: 200,000/mo (≈ 5/minute)
  • Medium: 1,000,000/mo (≈ 20/minute)
  • Large: 10,000,000/mo (≈ 230/minute or 4/second)

Search by meaning

When enabled, allows your users to enter non-exact search terms and get back results based on the semantic meaning of their query.

Search by meaning is only available on certain plans:

  • Free: ❌
  • Small: ❌
  • Medium: ✅
  • Large: ✅

Document embedding

Search by Meaning converts your documents into vector embeddings which represent the semantic meaning of the text, allowing your users to search by what the text means, not just what it says.

To ensure fair use, we impose a limit on this process that we predict the vast majority of customers will not reach. We calculate the number of tokens embedded using OpenAI’s tiktoken library configured for the cl100k_base encoding.

The number of tokens embedded is limited to 8,000 tokens per page included with your pricing plan, which is approximately 5,200 words per page. This is a total, so embeddings will be paused once the limit has been reached and will resume automatically once you drop below the limit on a rolling 30-day window.

This limit applies to embedding documents but not search queries. If you’ve hit your document embedding limit, your users can still make hybrid search queries.

  • Free: ❌
  • Small: ❌
  • Medium: 200 million total tokens (25,000 pages × 8,000 tokens/page)
  • Large: 400 million total tokens (50,000 pages × 8,000 tokens/page)

Please note that:

  • The text content of your pages is chunked with a 15% overlap. These limits include the tokens required to include overlap.
  • If a page has already been indexed, reindexing it will require vector embeddings to be generated again, which will count towards this limit.

If you consistently hit your document embedding limit every month, consider increasing your refresh interval or disabling automatic refreshing altogether.

Number of sources

There is no hard limit on the number of sources you can create. However, there may be a limit to how many sources you can query at once:

  • If Search by Meaning is enabled, you can include a maximum of 62 sources in one query.
  • If Search by Meaning is disabled, there is no hard limit on the number of sources you can include in one query.

This is a technical limitation, so we cannot raise it for certain customers.

Response body size

The maximum size of a response body from your server must not exceed 2 MB (2,097,152 bytes).

Text content length

The text content of a page, which includes its title, description, and content, is truncated after 200,000 characters.

Crawl speed

You can set crawls to occur at a rate between 1 and 300 pages per minute.

Embedding speed

If you have Search by Meaning enabled, Easysearch will embed 32 chunks, which contain up to 200 tokens each, every 2 seconds.

Search query length

All search queries, including from prebuilt search pages and the JSON API, are truncated to 150 characters.