SearchApi review, pricing and limits

Pick SearchApi for work where the real results page is the point rather than a summarised answer: rank tracking, AI Overview monitoring, Maps and Shopping research, ad library sweeps. A sample call in its docs took 1.25 seconds, so plan one to two seconds per search. Choose Linkup or Exa when you want ranked content you can quote straight away.

SearchApi is a REST search engine results API. One GET to /api/v1/search with an engine name and a query returns the parsed results page as JSON: organic results, knowledge graph, answer boxes, related searches and pagination. Its docs listed 59 engine endpoints on the day of this check, from Google web and AI Mode to Maps, YouTube, Amazon and four ad libraries. Any agent that can call HTTP can use it.

Maker: SearchApi · Protocol: REST · Auth: api key

Compatible agents: Claude (tool use), ChatGPT and OpenAI function calling, Gemini, LangChain and LlamaIndex tools, n8n and Make HTTP nodes, Orthogonal gateway (listed provider), any agent that can call a REST endpoint

Required runtime: Any HTTP client (curl, Python requests, Node fetch), A SearchApi account and API key

About SearchApi

SearchApi is a search engine results API. An agent sends one GET request with an engine name and a query and gets back the page a person would see, parsed into JSON: organic results with position, title, link and snippet, the knowledge graph, answer boxes, People Also Ask, inline videos, related searches and pagination. Its documentation covers Google web search, Google AI Mode and AI Overview, Gemini, Bing Copilot and Perplexity, Google Maps, Shopping, Trends, Scholar, News, Jobs, Flights and Hotels, YouTube, Amazon, eBay, Walmart, TikTok, and the Meta, LinkedIn, Reddit and TikTok ad libraries. That breadth is what makes it a skill rather than a search tool: it is the capability an agent plugs in when the job is to see what a real search engine returns, not to get a summarised answer.

It is a plain REST API, no SDK required. The call is a GET to the /api/v1/search endpoint with an engine name and a query string, and the API key as a query parameter or an Authorization: Bearer header. Localisation comes from location (a canonical place name the service maps to Google's own encoding), gl for country, hl for language, device for desktop, mobile or tablet, and latitude and longitude for exact targeting. Page selects the results page; since Google phased out the num parameter in September 2025 a Google web search returns ten results per page, and the separate Google Rank Tracking engine covers the top hundred. Every stored search keeps a JSON and an HTML copy at its own URL. Budget one to two seconds per call.

Any agent framework that can make an HTTP call can use it: Claude, ChatGPT and Gemini tool definitions, LangChain and LlamaIndex, automation platforms with an HTTP node, or the Orthogonal gateway, which lists SearchApi among its providers. Typical jobs are rank tracking for an SEO agent, checking what Google's AI Mode or AI Overview says about a brand, pulling Maps or Shopping results for local and price research, and reading ad libraries for competitive monitoring. Compare it with Linkup, which returns an LLM-ready answer with sources rather than the raw results page, and with Exa and Tavily, which are search engines built for agents rather than a window onto Google. Guides that place it: best AI search and retrieval infrastructure and best AI search tools.

Pricing is per successful search, with a free allowance on signup and no card. Monthly plans run from the Developer plan to Octo 1M at one million searches, with the per-thousand price falling as volume rises; every paid plan carries a stated uptime SLA, and Production and above add team management and the vendor's Legal Protection Guarantee, which it describes as accepting legal responsibility, under U.S. law, for how it collects and parses public search results. Full figures are in the cost question below. Other options for the same slot sit in the skills directory; if the question is which search layer fits your stack, Smart Match will ask.

Key Features

• 59 engine endpoints: Google web, AI Mode, AI Overview, Maps, Shopping, Trends, Scholar, News, Jobs, Flights, Hotels, YouTube, Amazon, eBay, Walmart, TikTok and four ad libraries, all through one request shape (docs navigation on the day of this check).
• Parsed JSON, not HTML: organic_results with position, link and snippet, knowledge_graph, answer_box, discussions_and_forums, inline_videos, related_searches and pagination come back as named blocks.
• Location targeting built in: A location name is mapped to Google's own encoding for you; gl, hl, device, and latitude and longitude give exact control.
• Pay per successful search: Failed requests are not billed; each completed search is kept with JSON and HTML copies at their own URLs.
• Legal Protection Guarantee: On Production plans and above the vendor states it accepts legal responsibility for how it collects and parses public results, with up to $2M of coverage under U.S. law.
• Resolved result links: link=resolved follows Google's encrypted redirect links in AI Overview and People Also Ask to their destination URLs, at the cost of some latency.

Use Cases

• Rank tracking for an SEO agent: Query Google for a keyword set with location and device fixed, read organic_results positions, and store the deltas week over week; use the Rank Tracking engine where the top hundred matters.
• AI answer monitoring: Pull the Google AI Mode and AI Overview blocks for brand and category queries and diff the cited sources against your own pages.
• Local and price research: Read Google Maps results for a service in a city, or Google Shopping and Amazon results for a product, into a comparison table an agent can reason over.
• Ad library sweeps: List a competitor's live ads across Meta, LinkedIn, Reddit and TikTok on a schedule and summarise what changed.

Install

curl "https://www.searchapi.io/api/v1/search?engine=google&q=best+ai+search+api&location=New+York,United+States&api_key=YOUR_API_KEY"

Requirements

• A SearchApi account (free allowance on signup, no card)
• An API key from the dashboard, sent as api_key or Authorization: Bearer
• An HTTP client; no SDK is required

Actions

Search

Run one search on a named engine and return the parsed results page as JSON.

curl -H "Authorization: Bearer YOUR_API_KEY" "https://www.searchapi.io/api/v1/search?q=electric+bikes&engine=google_ai_mode&location=Austin,Texas,United+States&gl=us&hl=en&device=mobile&page=1"

• engine (string) — required: Which engine to query, e.g. google, google_maps, google_ai_mode, youtube, amazon, bing. Any engine listed in the docs.
• q (string) — required: The search terms. Operators such as site:, inurl:, intitle:, AND and OR are supported on Google.
• api_key (string) — required: Sent either in the api_key query string or as a Bearer token in the Authorization header.
• location (string): Canonical place name; the most popular match is used (e.g. New York,United States). Cannot be combined with uule or coordinates.
• uule (string): Google-encoded location, for exact control instead of location.
• latitude (number): Decimal degrees, used with longitude; gl is inferred from the coordinates.
• longitude (number): Decimal degrees, used with latitude.
• gl (string): Country code of the search.
• hl (string): Interface language of the search.
• device (string): desktop, mobile or tablet.
• page (number): Results page to return; Google returns ten results per page.
• link (string): raw keeps Google's encrypted redirect links; resolved follows each one to its destination URL (adds latency).

How to Invoke

One HTTP GET to /api/v1/search with engine, q and the localisation parameters; the API key goes in the api_key query parameter or an Authorization: Bearer header. The agent reads the JSON blocks it needs (organic_results, knowledge_graph, answer_box, pagination).

Pricing

100 free requests on signup, no card (searchapi.io/pricing, read 11 October 2026). Monthly plans, billed only on successful searches. Developer: $40/mo, 10,000 searches. Production: $100/mo, 35,000 searches, adds team management and the Legal Protection Guarantee. BigData: $250/mo, 100,000 searches. Scale: $500/mo, 250,000 searches. Octo 500K: $900/mo. Octo 1M: $1,500/mo, 1,000,000 searches. All paid plans state a 99.9% SLA. Enhanced speed is an option on every plan; its price is not on the plan cards.

Strengths

• One request shape across every engine it lists, so an agent that learned the Google call can read Maps, YouTube or an ad library without new code.
• Returns the page as users see it, including AI Mode and AI Overview blocks, which answer-engine monitoring needs and summarising search APIs cannot give.
• Per-success billing with a free allowance to start and a per-thousand price that falls as volume rises.

Weaknesses

• Ten results per page on Google since the num parameter was removed in September 2025; deeper rankings mean more pages or the separate Rank Tracking engine.
• A sample call in the docs took 1.25 seconds, so a research loop of dozens of searches is tens of seconds; this is not a sub-second lookup.
• Output is the results page, not an answer: the agent still has to read snippets or fetch pages, where Linkup or Exa hand back ranked content directly.

View the official SearchApi skill page