Best of · Web, data & documents

Best web search APIs for AI agents

The 10 highest-scoring of 16 web search APIs on the Anchor benchmark, with a pick for each need and where each one falls short. Scores come from public evidence, re-checked as vendors change.

  • 16 ranked
  • 3 agent-ready
  • 2 accept x402
  • 14 hosted endpoints
  • Updated 9 October 2026

Top three

Picks by need

Worked out from the scores, prices and facts, so they change when the research does.

Highest score overall

Tavily API + MCP BB

BB, 76.8/100 on the benchmark.

Also You.com APIs, BB, 76.5/100.

Schema & documentation

Parallel Search and Task APIs BB

94/100 on schema & documentation, against 89 for the overall leader.

Agent ergonomics

Exa API + MCP B

96/100 on agent ergonomics, against 81 for the overall leader.

Security & auth

Cloudflare Web Search API C

71/100 on security & auth, against 52 for the overall leader.

Maintenance & community

Exa API + MCP B

88/100 on maintenance & community, against 85 for the overall leader.

Transparency & trust

SerpApi B

84/100 on transparency & trust, against 60 for the overall leader.

Lowest paid price per 1,000 requests

Cloudflare Web Search API C

$0.25 per 1,000 requests, the lowest of the 12 listings here with a paid price in this unit (free allowances aside).

Also Serper, $0.30 per 1,000 requests.

Paying per call with no account (x402)

Linkup B

accepts x402, $0.01 a call.

Also Exa API + MCP, accepts x402, $0.007 a call.

The review panel's favourite

You.com APIs BB

3.8/5 from 8 panel reviews.

The shortlist

#ToolGradeBest forPriceWhere
1 Tavily API + MCP
Tavily
BB 76.8 An agent that needs search plus page text in one call and has no account yet. $7.50 / 1k req hosted and local
2 You.com APIs
You.com
BB 76.5 An unattended agent that needs search, page text and cited answers and wants to pay per call without an account. $5 / 1k req hosted and local
3 Parallel Search and Task APIs
Parallel Web Systems
BB 74 Agents that mix quick search with long research or enrichment jobs and want one vendor and one bill. $1 / 1k req hosted
4 Linkup
Linkup
B 68.9 Agents that want a cited answer or schema-shaped output from one search call at a fixed price, and for European buyers who want a French vendor with zero retention on request. $0.01 / call hosted and local
5 Brave Search API + MCP
Brave
B 68 Agents that want an independent, non-Google index at a flat price, or that want page text in one call through LLM Context. $5 / 1k req hosted and local
6 Exa API + MCP
Exa
B 66 Agents that search in natural language and want short highlights or page text in the same call, and for keyless or wallet-paid use. $0.007 / call hosted and local
7 SerpKite
SerpKite (Anastasia Klimova, trading as SerpKite)
B 64.8 Agents that want Google results and the top pages as Markdown in one call at a low per-1,000 price, with spend caps per key. $1 / 1k req hosted
8 Valyu
Valyu
B 64.6 Research agents that need SEC filings, papers, patents, clinical trials or market data alongside the web, with a per-query price cap. $1.50 / 1k pages hosted
9 Grep AI
Parcha Labs, Inc.
B 64.4 Agents that hand off a long research or due-diligence job and want a cited report or structured fields back, including business verification and sanctions screening through built-in agents. $200 / mo hosted
10 TinyFish Search
TinyFish
B 63.2 Agents that need ranked snippets and URLs at no cost and modest volume, with news and research-paper modes. Freemium hosted

6 more are ranked in the full table.

How to choose

  1. Snippets, full text or an answerCheck whether results come back as snippets, full page text or a written answer, since a snippet may need a second fetch before the agent can ground a claim.
  2. Freshness of the indexCheck how recently the index updates and whether you can filter by date, since a stale result can make an agent cite a page that has already changed.
  3. Cost per query and credit expiryWork out the cost per query at your expected volume and check whether credits expire, since credits that lapse unused raise the real cost of each query.
  4. Retention and reuse of resultsCheck how long results and queries are logged and whether you may store or cache returned text, since the terms can limit what your agent is allowed to keep.

Each one in detail

#1

Tavily API + MCP

BB 76.8/100

Tavily's REST API for web search, URL extraction, site mapping, crawling and cited research reports, with the official MCP server hosted at mcp.tavily.com or run locally from npm.

Verdict Keyless search and extract, plus a free 1,000 credits a month with no card. x402 covers advanced search only, not extract, map, crawl or research.

Choose it for An agent that needs search plus page text in one call and has no account yet.

Strengths

  • Keyless search and extract, plus a free 1,000 credits a month with no card
  • x402 endpoint for advanced search at $0.01 in USDC on Base
  • Public OpenAPI, llms.txt and an error table that tells plan limits (432) from pay-as-you-go limits (433)

Weaknesses

  • x402 covers advanced search only, not extract, map, crawl or research
  • Hosted MCP docs lead with ?tavilyApiKey= in the URL query string
  • No readOnlyHint or destructiveHint on the MCP tools, and the npm server (0.2.22, 5 August 2026) trails the GitHub source

Price $7.50 / 1k reqAuth OAuth or keyx402 partialhosted and local

Full assessment

#2

You.com APIs

BB 76.5/100

Web Search, Contents, Answer, Research and Finance Research APIs, plus a hosted MCP server with six tools.

Verdict x402 and MPP on Web Search and Finance Research, with no account needed. Two API hosts. Answer and Research return 'Missing Authentication Token' on ydc-index.io.

Choose it for An unattended agent that needs search, page text and cited answers and wants to pay per call without an account.

Strengths

  • x402 and MPP on Web Search and Finance Research, with no account needed
  • $100 free credit with no card, plus a keyless MCP profile for search (100 queries a day)
  • Public OpenAPI, llms.txt and an error reference with guidance per code

Weaknesses

  • Two API hosts. Answer and Research return 'Missing Authentication Token' on ydc-index.io
  • No public changelog or deprecation notices
  • Zero Data Retention only for Web Search and Answer, and only on enterprise agreements

Price $5 / 1k reqAuth OAuth or keyx402 partialhosted and local

Full assessment · Against #1, Tavily API + MCP

#3

Parallel Search and Task APIs

BB 74/100

Parallel's web APIs for agents.

Verdict Fast-mode search at $1 per 1,000 with 10 excerpted results, and Extract at $1 per 1,000 URLs. Search defaults to advanced mode at $5 per 1,000 when mode is left out.

Choose it for Agents that mix quick search with long research or enrichment jobs and want one vendor and one bill.

Strengths

  • Fast-mode search at $1 per 1,000 with 10 excerpted results, and Extract at $1 per 1,000 URLs
  • Task runs billed only on success, and documented limits of 600 searches and 2,000 task creations a minute
  • Keyless hosted Search MCP, with Bearer and OAuth endpoints for higher limits

Weaknesses

  • Search defaults to advanced mode at $5 per 1,000 when mode is left out
  • SDKs retry Task creation on 429 and 5xx with no idempotency key
  • x402 and MPP only through parallelmpp.dev at a flat $0.01, not on api.parallel.ai

Price $1 / 1k reqAuth OAuth or keyx402 partialhosted

Full assessment · Against #1, Tavily API + MCP

#4

Linkup

B 68.9/100

Web search API for agents with four depths (flash, fast, standard, deep) and three output types (ranked sources, a cited answer, or JSON matching your schema).

Verdict Per-call prices from $0.005, and no credit taken for errors, missing parameters or empty results. 429 covers both exhausted credit and rate limiting, with no Retry-After.

Choose it for Agents that want a cited answer or schema-shaped output from one search call at a fixed price, and for European buyers who want a French vendor with zero retention on request.

Strengths

  • Per-call prices from $0.005, and no credit taken for errors, missing parameters or empty results
  • x402 on /v1/search and /v1/fetch at $0.01 with no account
  • Sources, a sourced answer or structured JSON from the same search call

Weaknesses

  • 429 covers both exhausted credit and rate limiting, with no Retry-After
  • MCP configs document ?apiKey= in the URL and an apiKey= argument for stdio
  • 10 queries a second per organisation by default

Price $0.01 / callAuth OAuth or keyx402 yeshosted and local

Full assessment · Against #1, Tavily API + MCP

#5

Brave Search API + MCP

B 68/100

REST API over Brave's independent index with web, news, image, video and local search, an LLM Context endpoint with pre-extracted page content and an Answers endpoint with citations.

Verdict x402 proxy sells web, LLM Context, news, video, image and local search at $0.005 a call, and a failed search isn't charged. The status page marks the Search API "not monitored" and shows no incident history.

Choose it for Agents that want an independent, non-Google index at a flat price, or that want page text in one call through LLM Context.

Strengths

  • x402 proxy sells web, LLM Context, news, video, image and local search at $0.005 a call, and a failed search isn't charged
  • $5 per 1,000 searches at 50 QPS on the self-serve Search plan
  • SOC 2 Type II, a HackerOne bounty and a valid security.txt

Weaknesses

  • The status page marks the Search API "not monitored" and shows no incident history
  • A card is required even for the $5 monthly credit
  • No public OpenAPI for the keyed API and no API changelog

Price $5 / 1k reqAuth OAuth or keyx402 partialhosted and local

Full assessment · Against #1, Tavily API + MCP

#6

Exa API + MCP

B 66/100

Exa's REST API over its own web index, with the official MCP server in front of it.

Verdict x402 on /search and /contents in USDC on Base or Solana, so an agent can pay per call with no account. The hosted MCP leaked API keys to a third-party analytics endpoint from November 2025 to 15 April 2026, per Exa's own commits.

Choose it for Agents that search in natural language and want short highlights or page text in the same call, and for keyless or wallet-paid use.

Strengths

  • x402 on /search and /contents in USDC on Base or Solana, so an agent can pay per call with no account
  • Two default MCP tools with about 1,250 characters of descriptions, and every tool marked read-only and idempotent
  • Keys can be minted by API with their own rate limit and budget

Weaknesses

  • The hosted MCP leaked API keys to a third-party analytics endpoint from November 2025 to 15 April 2026, per Exa's own commits
  • Agnost usage analytics in the MCP server with no disclosure or opt-out
  • The privacy policy says query data trains Exa's models and gives no retention period

Price $0.007 / callAuth OAuth or keyx402 yeshosted and local

Full assessment · Against #1, Tavily API + MCP

#7

SerpKite

B 64.8/100

Google SERP API for AI agents from SerpKite, a sole trader in Ontario. It returns live Google results and 10 other Google verticals as JSON or Markdown, fetches pages as Markdown, and runs a remote MCP server with 14 tools.

Verdict Google SERP API with an OpenAPI 3.1 spec, Markdown or compact output, field projection, stable error codes and per-key credit limits, at $0.60 to $1.00 per 1,000 on packs that don't expire. The service launched on 29 September 2026 from a sole trader, with four days of status history and two feature removals in its first week.

Choose it for Agents that want Google results and the top pages as Markdown in one call at a low per-1,000 price, with spend caps per key.

Strengths

  • OpenAPI 3.1 spec, llms.txt, llms-full.txt and Markdown twins of every docs page
  • format=markdown or compact, fields projection and an X-Tokens-Estimate header size responses for a model
  • Many revocable keys per account, each with an optional monthly credit limit, plus an account spend cap

Weaknesses

  • Launched 29 September 2026 by a sole trader, with status monitoring only since 2 October
  • AI Overview, AI Mode and Lens were removed within four days of v1.0.0, each without advance notice
  • GET requests accept the key as a ?api_key= query parameter, and keys have no scopes

Price $1 / 1k reqAuth API keyx402 nohosted

Full assessment · Against #1, Tavily API + MCP

#8

Valyu

B 64.6/100

Search API that covers the web plus specialised sources (arXiv, PubMed, SEC filings, market data, patents, clinical trials, genomics).

Verdict Web plus academic, financial, patent and life-science sources behind one search call. No published rate limits and no Retry-After or backoff guidance.

Choose it for Research agents that need SEC filings, papers, patents, clinical trials or market data alongside the web, with a per-query price cap.

Strengths

  • Web plus academic, financial, patent and life-science sources behind one search call
  • Public OpenAPI 3.0.3 with 26 paths, tight parameter constraints and a 206 partial-success code
  • Spend-capped, rotatable keys from the CLI, and scoped management keys

Weaknesses

  • No published rate limits and no Retry-After or backoff guidance
  • Web results cost $1.50 per 1,000 results retrieved, so a 10-result search is $0.015
  • MCP docs lead with ?valyuApiKey= in the URL query string

Price $1.50 / 1k pagesAuth OAuth or keyx402 nohosted

Full assessment · Against #1, Tavily API + MCP

#9

Grep AI

B 64.4/100

Grep is a hosted platform from Parcha Labs for building and running research agents that return cited reports and structured output. Outside agents reach it through a REST API, a remote MCP server, an A2A endpoint and two SDKs.

Verdict The public API has a 126-operation OpenAPI document, scoped OAuth and API keys, idempotency keys with documented replay rules and a per-run cost cap. No rate limit is published, the MCP server lists 82 tools, and the pricing page limits API access to the $200 a month Pro plan and above.

Choose it for Agents that hand off a long research or due-diligence job and want a cited report or structured fields back, including business verification and sanctions screening through built-in agents.

Strengths

  • Public OpenAPI document with 126 operations, llms.txt and a Markdown twin of every developer page
  • OAuth with five scopes, plus API keys that can be minted, scoped and revoked through the API
  • Idempotency-Key on run creation, with body-hash matching, a 24-hour replay window and named 409 codes

Weaknesses

  • No rate limit numbers in the reviewed documentation, and no published SLA
  • The MCP server card lists 82 tools, 31 of them deprecated aliases, with one-line descriptions and no annotations
  • The pricing table marks API access as Pro and Ultra only, from $200 a month

Price $200 / moAuth OAuth or keyx402 nohosted

Full assessment · Against #1, Tavily API + MCP

Head to head

All 129 comparisons in this category

Questions

What are the highest-rated web search APIs for AI agents?

Tavily API + MCP has the highest benchmark score of the 16 ranked web search APIs, 76.8 (BB). You.com APIs is second with 76.5 (BB).

How many web search APIs are agent-ready?

3 of the 16 ranked here grade BB or better, the bar for agent-ready on the Anchor benchmark.

Which web search APIs accept x402 payments?

Linkup and Exa API + MCP. An agent can pay these per call in USDC with no account.

Which of these web search APIs is cheapest?

By published paid prices, Cloudflare Web Search API, at $0.25 per 1,000 requests, the lowest of the 12 listings here with a paid price in this unit (free allowances aside). Plans, volume tiers and free allowances change the sum, so check the listing's price table.

How is this list ranked?

By the Anchor benchmark score out of 100, a weighted mean of the scored categories minus deductions for negative events, from public evidence re-checked as vendors change. Listings cannot pay for a place. The latest assessment behind this page is from 9 October 2026.

How this list is made

The order is the Anchor benchmark score, the same number as on each listing and in the top list. Each listing is graded from public evidence against the benchmark checklist, and the picks above are worked out from those grades, prices and facts. No listing pays for its place, and paid audits or listing help never change a score.

Full ranked table · 129 head-to-head comparisons · Best tools in every category

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.