Does it work with other engines?

Yes. include_content fetches the top results whichever engine answered the search.

How long does it take?

The slowest of the N pages bounds the response, so use a client timeout of at least 60 seconds, or queue the request with POST /v1/batches at half price.

What if I already have the URLs?

Use the Web Extract API for up to 20 URLs per call, or the Website Crawler API for a whole site section.

Are highlights made by an LLM?

No. They're ranked with BM25, a lexical ranker. Same input, same passages, no model cost.

include_content + highlights

Search with page content

Snippets are rarely enough to answer a question. include_content fetches the top results' pages as Markdown in the same /v1/search call; highlights returns only the passages that match your query.

Get a free API key Page content reference

How it works

Top 1–5 result pages as Markdown, or ranked passages, in one call.

The usual RAG loop is a search followed by N page scrapes. With include_content: 1…5, SerpKite runs the search, fetches the top N organic results, converts each page to clean Markdown and puts it into results[].content. One request, one bill, one timeout to manage.

highlights: true replaces each page with its 3 passages most relevant to q, about 600 characters each, split along the page's headings and ranked with BM25. It's deterministic, needs no model and costs nothing extra, and it keeps prompts small.

Pages built by JavaScript are rendered in a browser when there's time. Pages that can't be fetched aren't billed, and the unused part of the reservation is refunded.

Search + read

cURL

TypeScript

Python

curl https://api.serpkite.com/v1/search \
  -H "Authorization: Bearer $SERPKITE_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{"q":"how does retrieval augmented generation work","include_content":3,"highlights":true}'

A result with highlights · illustrative

{
  "position": 1,
  "title": "Async in depth",
  "link": "https://docs.example.org/async",
  "domain": "docs.example.org",
  "published_at": "2026-09-30",
  "highlights": [
    {
      "text": "An async runtime drives futures to completion by polling them…",
      "score": 7.412,
      "heading": "Async in depth > Runtimes"
    }
  ]
}

What it costs

What it costs
Request	Credits
`/v1/search`	1
`/v1/search` + `include_content: 3`	up to 4
`/v1/search` + `include_content: 5`	up to 6
`include_content: 3` + `highlights: true`	up to 4 (highlights are free)
`include_content: 3` via `POST /v1/batches`	up to 2

Why it matters

What you get

One round trip

Search and page reads in a single call, with one latency budget instead of N.

Passages, not pages

highlights returns the 3 best passages per page with a score and heading path.

Pay for pages that load

Timeouts, blocked and empty pages aren't billed; the unused reservation is refunded.

Public pages only

Fetched as a logged-out visitor sees them. Nothing behind a login or paywall.

FAQ