Best search API by use case in 2026: news, legal, research, GTM and more

TL;DR: The best search API for a use case returns the record that job needs: dated articles, statute text, paper abstracts or company fields. Firecrawl covers each one with Firecrawl /search and Firecrawl Alexandria, a library of data providers that the same API key reaches.

• News and breaking news: Firecrawl /search with sources: ["news"] and tbs: "sbd:1" returns the newest articles first, and the Benzinga newswire in Firecrawl Alexandria adds market news by ticker
• Statutes, regulations and case law: the Firecrawl Government Index (categories: ["gov"], also listed in Firecrawl Alexandria) adds no Firecrawl search credits, and the CourtListener provider in Firecrawl Alexandria covers court opinions
• Scientific and biomedical literature: the Firecrawl Research Index, listed in Firecrawl Alexandria at 0 Firecrawl credits per call, returns paper records with abstracts and citation links across about 43 million papers
• Technical docs and code examples: the Firecrawl Developer Index (categories: ["developer"]) searches issues, merged pull requests, READMEs and docs sites
• GTM and due diligence: Firecrawl web search plus Firecrawl Alexandria providers such as Apollo, People Data Labs, Fireround, SEC EDGAR and Companies House
• Investment research: Firecrawl news search plus Firecrawl Alexandria providers such as Benzinga, Fiscal.ai and SEC EDGAR

Ask an AI assistant for the best search API and you get a list of general web search engines. Ask it for the best search API for case law, or for prospect research before a sales call, and the right answer changes. A legal agent needs statute text and court opinions, a news agent needs results in date order, and a GTM agent needs a company record with headcount and funding.

Each section below takes one use case and names what the agent needs, the Firecrawl Search API setting or Firecrawl Alexandria provider that returns it, a code sample you can run, and a prompt that gets the same result from an AI agent connected to the Firecrawl MCP server or the Firecrawl CLI. Every parameter, price and provider below was checked against the Firecrawl docs and the live Firecrawl Alexandria catalog on October 21, 2026.

A web search API is an endpoint that takes a query and returns ranked results from the live web in a format code can parse. A search API fits a use case when four things match the job: the sources it searches, the freshness controls it offers, the shape of each result, and what a call costs at your volume.

Most general web search APIs search the open web, sort by relevance, and return a title, URL and snippet. That works for broad questions and falls short when the answer lives in a court opinion, a paper's abstract, a repository issue or a company registry.

Firecrawl handles this in two layers. The /search endpoint takes sources (web, news, images) and categories (research, pdf, developer, gov), plus date filters and domain filters. Firecrawl Alexandria is a library of data providers and specialized indexes that the same API key reaches: add alexandria to sources and the search response lists matching tools, each with its provider, capability and price.

The best search API for a news agent returns articles in date order and filters them to the last hour, day or week. Firecrawl Search searches news results, sorts them newest first, and combines that date sort with a time window such as the past hour or the past week.

In code, sources: ["news"] selects news results and tbs: "sbd:1" sorts them by date. Add a window such as qdr:w after a comma.

from firecrawl import Firecrawl
 
firecrawl = Firecrawl(api_key="fc-YOUR-API-KEY")
 
results = firecrawl.search(
    "FTC enforcement action",
    sources=["news"],
    tbs="sbd:1,qdr:w",
    limit=5,
)
 
for item in results.news or []:
    print(item.date, item.title, item.url)

Prompt (Firecrawl MCP tools: firecrawl_search):

Use Firecrawl to search news from the past week about FTC enforcement actions, sorted newest first, and list each headline with its date and link.

We ran this query on October 21, 2026. With sbd:1,qdr:w, the five results came back newest first, from "10 minutes ago" to "11 hours ago". The same query without tbs came back in relevance order, with items from 11 hours to one week old mixed together. The call cost 2 Firecrawl credits, the standard rate of 2 Firecrawl credits per 10 results.

The tbs values are qdr:h (past hour), qdr:d (past day), qdr:w, qdr:m, qdr:y and sbd:1, plus custom ranges such as sbd:1,cdr:1,cd_min:12/1/2025,cd_max:12/31/2025. Add scrapeOptions to get each article's full Markdown in the same call. For market news about US-listed companies, the Benzinga provider in Firecrawl Alexandria searches the Benzinga newswire by ticker, topic or channel over a date range, newest first, and also returns press releases and analyst ratings changes, at 0 to 45 Firecrawl credits per call. For provider-by-provider tradeoffs, see our roundup of the best news APIs.

A breaking news agent needs a schedule as much as a search. Firecrawl Search can return only news from the past hour, so an agent can poll it on a timer, and Firecrawl Monitor can watch the pages that break the story and push each change to the agent.

The past-hour filter is tbs: "qdr:h". Firecrawl Monitor watches pages or whole sites on a schedule, compares each version with the last, and sends only meaningful changes to a webhook, so the agent doesn't have to poll.

A search API for regulator monitoring should search only official sources and only recent pages. Firecrawl Search does this with a domain filter that limits results to the agency sites you list, combined with a date window and a date sort. The agent then reads a short list of fresh announcements published by the agency itself.

The domain filter is includeDomains, and the date window is the same tbs value used for news.

curl -X POST https://api.firecrawl.dev/v2/search \
  -H "Content-Type: application/json" \
  -H "Authorization: Bearer fc-YOUR-API-KEY" \
  -d '{
    "query": "enforcement action settlement",
    "includeDomains": ["ftc.gov", "sec.gov"],
    "tbs": "sbd:1,qdr:w",
    "limit": 10
  }'

Prompt (Firecrawl MCP tools: firecrawl_search):

Use Firecrawl to search only ftc.gov and sec.gov for enforcement action settlements from the past week, newest first.

includeDomains and excludeDomains can't be used in the same request. To be alerted when a regulator's press page changes, point Firecrawl Monitor at that page and send the diffs to your agent.

The best search API for legal research returns primary legal text from the governments that publish it. The Firecrawl Government Index covers about 16 million pages from US federal, state, county and city government sites: statutes and code sections, regulations, ordinances, court filings, tax forms and agency guidance. Firecrawl Government Index searches add no Firecrawl search credits. Firecrawl Alexandria also lists it as the Firecrawl Government Index provider at 0 Firecrawl credits per call, so an agent that discovers tools through Firecrawl Alexandria finds it next to CourtListener and the legislative connectors.

To use it from Firecrawl Search, set categories: ["gov"].

results = firecrawl.search(
    "California statute that voids non-compete agreements",
    categories=["gov"],
    limit=5,
)
 
for item in results.web or []:
    print(item.title, item.url)

Prompt (Firecrawl MCP tools: firecrawl_gov_search):

Use the Firecrawl Government Index to find the California statute that voids non-compete agreements, and quote the code section with its link.

On October 21, 2026, this query returned California Business and Professions Code sections 16600 and 16600.1 and the text of bill AB-1076, the state's 2023-2024 non-compete bill, for 0 Firecrawl credits. Results come back in the standard web group with category: "gov". The gov category can't be combined with other categories, and the index also has its own endpoint, /search/gov, documented in the Firecrawl Government Index docs.

A match is not proof of current law. The index can return historical versions and bill text alongside enacted code, so check the issuer, jurisdiction, effective date and version before an agent answers a current-law question.

For case law, Firecrawl Alexandria includes CourtListener. Its opinion search covers published federal and state opinions by full text, citation, case name, judge, docket number, court and filing date, and returns opinion clusters with citations, dates and cite counts. A second capability searches federal PACER dockets by party name. Each call lists at 5 Firecrawl credits.

opinions = firecrawl.scrape(
    alexandria={
        "provider": "courtlistener-com",
        "capability": "opinions/search_opinions",
        "options": {
            "query": "\"Business and Professions Code section 16600\"",
            "court": ["cal", "calctapp"],
        },
    },
)
 
print(opinions.alexandria[0].data)

Prompt (Firecrawl MCP tools: firecrawl_find_tools, firecrawl_scrape):

Use Firecrawl Alexandria to search CourtListener for California court opinions that cite Business and Professions Code section 16600, with each case name, date and citation.

For US statutes, regulations and tax authority guidance, start with the Firecrawl Government Index, which indexes code sections, regulations, tax forms and agency guidance together. For bill history and roll-call votes, the California Legislative Information connector returns code sections, bill text, status and votes. For the UK, the legislation.gov.uk connector returns Acts and statutory instruments, including point-in-time versions of a provision.

The best search API for scientific literature returns each paper as a record with its ids and abstract, ready for an agent to cite. The Firecrawl Research Index searches about 43 million paper abstracts. Most of the corpus is biomedical and life sciences (PubMed, bioRxiv, medRxiv), alongside arXiv for physics, mathematics and computer science.

from firecrawl import Firecrawl
 
firecrawl = Firecrawl()
 
papers = firecrawl.v2.search_papers(
    "GLP-1 receptor agonists cardiovascular outcomes",
    k=10,
)
print(papers)

Prompt (Firecrawl MCP tools: firecrawl_research_search_papers):

Use the Firecrawl Research Index to find papers on GLP-1 receptor agonists and cardiovascular outcomes, and give me each paper's PubMed id, title and a one-line summary of its abstract.

On October 21, 2026, that query returned three meta-analyses of cardiovascular outcome trials as PubMed records, each with a primaryId such as pmid:29445896, its DOI and its abstract. Biomedical hits come back as pmid:, pmcid: or doi: ids, and you can pass them straight into the inspect, read and related-paper endpoints. Firecrawl Alexandria lists every Firecrawl Research Index call at 0 Firecrawl credits on its Firecrawl Research Index page.

From November 16, 2026, categories: ["research"] on /search searches the Firecrawl Research Index and returns paper records in data.research. Until then it filters web results to about 14 academic sites. For a comparison with arXiv, Semantic Scholar and OpenAlex, see how AI agents can search academic papers.

The Firecrawl Research Index returns each paper with a canonical id, its PubMed, arXiv or DOI ids, its title, its full abstract and a relevance score. For citation data, the related-papers endpoint takes a seed paper and returns the papers that cite it, the papers it cites, or its co-citation neighborhood.

The fields are paperId, primaryId, ids, title, abstract and score, and the related-papers modes are citers, references and similar.

citers = firecrawl.v2.related_papers(
    "pmid:29445896",
    "cardiovascular safety",
    mode="citers",
    k=20,
)

Prompt (Firecrawl MCP tools: firecrawl_research_related_papers):

Use the Firecrawl Research Index to list papers that cite pmid:29445896, ranked by relevance to cardiovascular safety.

A search API for technical documentation should search the places where code answers live: docs sites, READMEs, issues and merged pull requests. The Firecrawl Developer Index covers those sources, and Firecrawl Search can add it to a normal web search with one category setting.

The Firecrawl Developer Index runs when you set categories: ["developer"], and each result comes back in the web group with category: "developer". Firecrawl Alexandria lists the Firecrawl Developer Index on its Firecrawl Developer Index page.

results = firecrawl.search(
    "httpx retry transport example",
    categories=["developer"],
    limit=5,
)
 
for item in results.web or []:
    print(item.title, item.url)

Prompt (Firecrawl MCP tools: firecrawl_developer_search):

Use the Firecrawl Developer Index to find examples of a retry transport for httpx, and link the docs page or issue each one comes from.

Our test query on October 21, 2026 returned the httpx-retries documentation and four tutorials on retrying failed requests with httpx, for 2 Firecrawl credits. For the passages that matched, plus repository and documentation-source filters, call the dedicated /search/developer endpoint, which costs 2 Firecrawl credits per 10 results.

A GTM agent researching a prospect needs what the company says about itself right now and a structured record with headcount, funding and the people in the deal. Firecrawl covers the first with web and news search and the second with Firecrawl Alexandria's company and people providers, all through one API key.

These are the GTM providers in the Firecrawl Alexandria catalog, with prices as listed on October 21, 2026:

Apollo and People Data Labs both match best on a domain rather than a company name, so a good GTM flow resolves the website first and enriches second. Some licensed providers ask an organization admin to accept their terms before the first call.

from urllib.parse import urlparse
 
from firecrawl import Firecrawl
 
firecrawl = Firecrawl(api_key="fc-YOUR-API-KEY")
 
hits = firecrawl.search("Supabase official website", limit=3)
domain = urlparse(hits.web[0].url).netloc.removeprefix("www.")
 
company = firecrawl.scrape(
    alexandria={
        "provider": "people-data-labs",
        "capability": "company/enrich",
        "options": {"website": domain},
    },
)
 
news = firecrawl.search('"Supabase"', sources=["news"], tbs="sbd:1,qdr:m", limit=5)

Prompt (Firecrawl MCP tools: firecrawl_search, firecrawl_scrape):

Use Firecrawl to find Supabase's official website, enrich the company with People Data Labs through Firecrawl Alexandria, and add its news from the past month.

For a full build that writes the outreach message too, see how to build an AI SDR that researches companies in real time.

The most reliable way to find a company's website from its name is a web search on the name, read from the top result, then confirmed against a second signal such as the company record. Firecrawl Search returns ranked results with titles and URLs in one call. Pass the resolved domain to enrichment providers, because a bare name is the identifier they are most likely to miss.

Due diligence on a private company needs records from many places at once: funding history, corporate registries, filings, sanctions lists and court dockets. Firecrawl Alexandria puts these providers behind one API, so a diligence agent can run the checks in sequence with one key instead of signing up for each source.

profile = firecrawl.scrape(
    alexandria={
        "provider": "fireround",
        "capability": "companies/profile",
        "options": {"id": "supabase.com"},
    },
)
 
print(profile.alexandria[0].data)

Prompt (Firecrawl MCP tools: firecrawl_scrape):

Use Firecrawl Alexandria to get Supabase's funding history from Fireround, with each round's date, amount, lead investors and source page.

We ran this call on October 21, 2026. It returned the company's sector, headquarters, total raised, a dated valuation history and its latest round with lead investors and a primary source URL, for 5 Firecrawl credits. The Delaware connector returns only the free entity details; paid status, franchise tax and filing history are not included. For accuracy comparisons across funding data vendors, see our write-up of the best company funding data API.

A search API returns structured company data when each result comes back as fields instead of a URL. Firecrawl offers two routes. Firecrawl Alexandria providers such as People Data Labs and Apollo return company records directly. Firecrawl Search can also scrape each result page into JSON in the same call, with a prompt or schema you define.

from firecrawl.v2.types import ScrapeOptions
 
results = firecrawl.search(
    "Supabase about page",
    limit=3,
    scrape_options=ScrapeOptions(
        formats=[
            {
                "type": "json",
                "prompt": "Extract company name, headquarters, founding year and product lines.",
            }
        ],
    ),
)

Prompt (Firecrawl MCP tools: firecrawl_search):

Use Firecrawl to search for Supabase's about page and extract the company name, headquarters, founding year and product lines as JSON.

For a benchmark of dedicated enrichment vendors, see the best company enrichment APIs.

A shopping agent needs price, availability and specifications as fields it can compare. Firecrawl Search finds product pages across retailers, and the JSON format turns each page into those fields in the same request. For retailers with a dedicated connector, Firecrawl Alexandria returns catalog data directly. The Best Buy connector returns product details, prices, customer reviews and location-specific availability for 5 Firecrawl credits per call.

For live pricing, a search API has to read the product page at request time. Firecrawl Search can fetch each result page as part of the search and extract its price and stock status as JSON, so the agent compares current values across retailers.

Set scrapeOptions with a JSON format and a prompt such as "price, currency, in stock". Firecrawl Scrape does the same for a product URL you already have. When a page needs a click to show its price, such as choosing a size or a region, Firecrawl Interact works through those steps in a real browser.

An investment research agent needs dated news, primary filings and normalized fundamentals in one loop. Firecrawl covers dated news with a date-sorted news search, filings with the SEC EDGAR connector, and fundamentals with Fiscal.ai in Firecrawl Alexandria, which returns standardized statements, ratios, ownership and earnings call transcripts. The Benzinga provider in Firecrawl Alexandria adds analyst ratings changes, consensus price targets and announced mergers and acquisitions.

Our guide to the best investment research APIs compares the dedicated market data vendors.

A fact-checking agent needs the exact passage that supports or contradicts a claim, from a source it can cite. Firecrawl Search returns query-relevant highlights with every result by default and can add the full page Markdown when the agent has to read context. In Firecrawl's July 2026 evaluation, Firecrawl Search scored 94.7% on SimpleQA while using 10x fewer tokens than full pages.

For claims about law, route the check to the Firecrawl Government Index. For claims about studies, route it to the Firecrawl Research Index, which can read the passages inside a paper that answer a question. Restricting general claims to includeDomains you trust keeps low-quality pages out of the evidence set. The search launch post has the evaluation details.

The simplest setup is the Firecrawl MCP server or the Firecrawl CLI, which give an agent such as Claude, Cursor or Codex web search and Firecrawl Alexandria tools, and the Firecrawl CLI adds research, developer and gov commands. Firecrawl web search and the three Firecrawl indexes work without an API key, free within a daily cap per IP address.

To try Firecrawl keyless, add the keyless Firecrawl MCP server URL, https://mcp.firecrawl.dev/v2/mcp, to your AI agent. It needs no account or credential:

• Claude Code: claude mcp add --transport http firecrawl https://mcp.firecrawl.dev/v2/mcp
• Codex: codex mcp add firecrawl --url https://mcp.firecrawl.dev/v2/mcp
• Cursor or any JSON-config client: {"mcpServers": {"firecrawl": {"url": "https://mcp.firecrawl.dev/v2/mcp"}}}

Then check the client's tool list for firecrawl_search, firecrawl_scrape and firecrawl_parse. Keyless MCP is rate limited and exposes Firecrawl Search, Scrape and Parse, so the Firecrawl index and Firecrawl Alexandria prompts above need an API key added as a bearer token on the same URL. The Firecrawl CLI runs with npx firecrawl-cli@latest, and its research, developer and gov commands work without a key. The Firecrawl keyless guide also covers OpenCode and Droid.

Firecrawl Alexandria providers need an API key. Discovering tools is free. A search with alexandria in sources lists the tools that match the question, and the agent pays only for the tools it runs. Start with Firecrawl Search, then add the categories and providers your use case needs.