Niches · Web search APIs

Search and get page text in the same call

Four or more companies' docs show a full fit.

7companies, full fit
6companies, fit with a limit
76companies carry the tag
14share of the category's work
8first-hand accounts

What counts as done

Each of the top results carries extracted page text or highlights of a controllable length (a max tokens or characters parameter).

Measured, for our own tests: the whole response fits in a few thousand tokens. No docs we read state these figures, so they don't count against a listing here.

How agents ask for it: find and read what the top pages say about X, give me the content, not only links, research X and quote the sources

Does the job 7

The docs show it meets the test above, and nothing we read says otherwise.

ListingWhy
Brave Search API + MCP B
Brave Software, Inc.
LLM Context returns extracted snippets per URL with maximum_number_of_tokens (1,024 to 32,768) and maximum_number_of_tokens_per_url; Web Search adds up to five extra_snippets per result. source
Exa API + MCP B
Exa Labs Inc.
contents.highlights or contents.text in the search request, each with maxCharacters; an informal test ranked Exa first for agent search. source
Parallel Search and Task APIs BB
Parallel Web Systems Inc.
Results are excerpts per URL, capped by max_chars_total; basic mode returns extended snippets. source
Perplexity Search API BB
Perplexity AI, Inc.
max_tokens_per_page and max_tokens cap the extracted snippet per result and for the response; a citation-checking user runs it at volume. source
SerpKite BB
Anastasia Klimova, trading as SerpKite
include_content fetches that many result pages as Markdown in the same call at 1 credit a page, with highlights ranked against the query and fields or compact format to trim the rest. source
Tavily API + MCP A
AlphaAI Technologies Inc. dba Tavily
include_raw_content returns page text per result and chunks_per_source caps it; include_usage reports credits. source
Valyu BB
Valyu.Network Ltd
Each result carries content, capped by response_length short, medium, large, max or a character count, with max_price bounding the request. source

Does the job, with a limit 6

It does the job, but something in the docs or a user's account will get in an agent's way. The limit is named in each row.

ListingWhy
Web Search Tool on Amazon Bedrock AgentCore BB
Amazon Web Services, Inc. (regional AWS entities by account location)
Each result's text field is a semantically extracted snippet, not a link only, but the only size control is maxResults; no token or character cap exists. source
Cloudflare Web Search API B
Cloudflare, Inc.
Ceramic returns descriptions up to 8,000 characters and Exa returns highlights as the description, but there is no size parameter. source
Kagi Search API C
Kagi Inc.
extract.count (1 to 10) replaces each snippet with the page Markdown in the same call, but it is billed at the Extract rate on top of the search and one user got no content back. source
Linkup B
Linkup Technologies SAS
Each result carries content (extracted text) and a snippet, but the reference has no character or token cap parameter. source
SpecStream B
SpecStream SAS
passages=true returns the text of the top pages closest to the request, adding about a second, with no length cap parameter. source
You.com APIs BB
SuSea, Inc., d/b/a you.com
extraction_mode highlights or full_page (markdown or html) returns text per result, but there is no token or character cap, only crawl_timeout and count. source

Couldn't check 1

The docs were closed to us, or silent. These carry no mark either way.

ListingWhy
DeepSeek Search and Fetch API C
Hangzhou DeepSeek Artificial Intelligence Co., Ltd.
Docs not opened (terms bar automated copying) and no account describes the need_content option.

Doesn't do this job 12

Listings in the category that do something else, with the reason.

ListingWhy
Arcmira B
Arcmira, Inc.
Passage text comes from transcripts, not web pages; the use case asks for extracted web page text.
Grounding with Bing Search B
Microsoft Corporation
No page text or highlights are returned, only the model's text with citation links. source
Fetch (MCP reference server) B
Model Context Protocol, a Series of LF Projects, LLC
No search.
Grep AI B
Parcha Labs, Inc.
No search endpoint; runs return a report.
Liner Search MCP E
Liner
Results carry a description only; no extraction or highlights parameter.
Mojeek Web Search API F
Mojeek Limited
Results carry a description only; no extraction option.
NewsCatcher B
NewsCatcher, Inc.
Snippets are the start of the publisher's description, cut to the plan's length; no extraction.
SearchAPI.io D
SearchApi, LLC
SERP snippets only; no page extraction.
SerpApi B
SerpApi, LLC
output=md is the SERP as Markdown, not page text.
Serper F
Serper
Snippets only; no page text in the response samples.
SERPHouse E
Patoliya Infotech LLP
SERP fields only; page text needs the separate Scraper API.
TinyFish Search B
Tiny Fish, Inc.
Search returns snippets; page text needs a second call to the Fetch API.

What stops agents finishing

  • extraction billed on top of the search without a cap (Kagi, Tavily advanced)
  • extracted text fills the context when no token cap is set (Perplexity)
  • SPA pages come back empty with no error (Tavily extract)

Around it

After: document-extraction.summarise, memory.store.

Where this comes from

Weight 14 of 100 in Web search APIs: 8 of 44 accounts chose a provider because text came back with the results (Kagi extract, Exa highlights, Perplexity, Tavily raw content, Linkup) or complained a provider only returned links

The same as data: /api/v1/use-cases/search.search-with-text.json, or the MCP tool get_use_case with id search.search-with-text. Other jobs in this category: /api/v1/niches?category=search.

Draft, read 2026-11-12 from public docs and first-hand accounts, not yet reviewed by a person. A mark says what the docs show, not what a run measured.

For companies

Do agents find, use and choose your tools?

An agent-readiness audit runs our probes, task suite and eight reviewer agents against your public and internal tools, and comes back with a scorecard, the transcripts of what failed, and a fix list in priority order. From $2,500, re-run included. We never take payment to move a rank. We do help companies earn one.