How Many Search Engines Are There?

By Bill Hartzer · Updated 19 August 2026

The short answer

There is no register of search engines, no licensing body, no standards organisation and no census. Anyone asserting a precise global figure has either counted something narrow and not said what, or made it up. What can be said honestly is that the answer differs by an order of magnitude depending on which of three questions you are actually asking.

How many websites will search something for you? Uncountable, and rising. Every site with a search box over its own content qualifies under a loose definition, as does every specialist database, every app-store search and every product catalogue.
How many general web search engines could you name and use today? Somewhere in the hundreds. This site profiles around 55 of the ones with a genuine claim on a reader's attention, and that is a curated selection rather than an exhaustive list.
How many organisations crawl and index the open web themselves, at scale? Roughly a dozen worldwide. This is the number that actually determines what you can find, and it is small enough to list by name.

The gap between the second and third figures is the entire subject. Most search engines are not search engines in the sense people assume. They are interfaces — branding, filtering, privacy handling and design placed in front of results bought from somebody else's index.

Why nobody can give you a real count

Four separate problems make a hard number impossible, and they are worth understanding because they also explain why published figures disagree so wildly.

The definition problem. Is a site-search box a search engine? Is a vertical that only indexes academic papers, or job listings, or code? Is an AI assistant that retrieves web pages before answering? Every published count silently picks an answer, and the answers are not compatible.

The instance problem. Some search engines are software rather than services. SearXNG is installed and run by whoever wants it, on their own terms, with their own choice of upstream engines enabled; its documentation describes support for up to 269 different search services and refers to roughly seventy well-maintained public instances at any given moment, alongside an unknown number of private ones. Are those one search engine, seventy, or several thousand? All three answers are defensible, which is a sign that the question is badly formed.

The measurement problem. The standard public source for search market share, StatCounter Global Stats, names only a handful of engines — Google, Bing, Yandex, Yahoo!, DuckDuckGo and Baidu — and puts everything else into an "Other" residual. So the only widely cited dataset in the field does not itself attempt a count, and cannot tell you whether the residual contains ten engines or ten thousand. It also measures page views on the sites it tracks, not people and not queries.

The brand-versus-engine problem. Dogpile, MetaCrawler and WebCrawler look like three search engines. They have shared one owner since around 2000 and run on the same infrastructure. Yahoo! Search has served Microsoft's results since 2010. Counting brands gives a number several times larger than counting the systems underneath, and almost every listicle counts brands.

The number that matters: independent indexes

Running a web-scale crawler and index means continuously discovering, fetching, storing and ranking billions of documents, and paying for the hardware, bandwidth and engineering to keep doing it. Very few organisations do it. Here is the honest global picture as of August 2026, with the qualifications that make it accurate rather than tidy.

The two that everything else consumes: Google and Microsoft Bing. Google's share is the reason. StatCounter put Google at 91.31% worldwide across all devices in July 2026 (gs.statcounter.com), and the United States district court that ruled on Google's search antitrust liability in August 2024 found it holds roughly 90% of general search and about 95% on mobile — a court finding rather than a vendor estimate.

Large national and regional indexes: Yandex (Russia), Baidu (China), Naver (South Korea), Seznam (Czech Republic, running its own index since 2005 after previously licensing one), Cốc Cốc (Vietnam) and Sogou (China). These are real crawlers with real indexes, but several are scoped rather than global — Cốc Cốc's own documentation says its crawler deliberately limits itself to Vietnamese content, and Naver's results page is dominated by Naver's own user-generated content rather than the open web it crawls. Sogou is a hybrid: it crawls Chinese pages itself while a 2016 arrangement had English results supplied by Bing and machine-translated, and whether that arrangement still stands is not documented publicly.

Independent Western indexes outside the duopoly: a very short list. Brave Search crawls and ranks its own index and has served web results with no Bing API calls since 27 April 2023 — at its 2021 launch roughly 87% of results came from its own index and the remainder from a third party, so this is a transition that actually completed. Mojeek, a small British company, has never resold anyone's results at any point in its history; its own About page gave 9 billion pages in 2025, against a Google index generally understood to run into the hundreds of billions. Marginalia Search, essentially one Swedish engineer, indexed around 300 million documents as of 2024 and deliberately optimises for non-commercial text rather than coverage. Perplexity published an engineering write-up in September 2025 describing an exabyte-scale index tracking over 200 billion unique URLs, a significant change from its early years as largely a Bing reseller with a summarisation layer.

Two more are being built. European Search Perspective, a 50:50 joint venture between Qwant and Ecosia, began serving results from its own index, Staan, in August 2025, starting in France. And Presearch has claimed since September 2025 to be running an index of its own, though as of August 2026 none of the artefacts a working crawler produces — a user-agent string, a published IP range, an opt-out mechanism for site owners, an index size — has appeared.

Count generously and you get about a dozen. Count strictly, requiring a global open-web crawl, and you get closer to six.

Everything else is a front end

Once you know how short the list is, the rest of the landscape becomes legible. The great majority of named search engines buy their results, and that is not a hidden scandal — several of them say so plainly and it is their actual product.

Deliberate resellers. Startpage exists to submit your query to Google and Bing anonymously on your behalf and return their results; the absence of an index is the point of the product. Swisscows serves Bing's results with a permanent adult-content filter. Yahoo! Search has been Bing-powered since 2010.
Hybrids. DuckDuckGo runs its own crawlers and its own instant-answer indexes but sources the ranked web links largely from Bing. Kagi crawls its own non-commercial and news indexes and blends them with anonymised calls into third-party indexes, selling the re-ranking and filtering as the product. Ecosia and Qwant blend partner feeds with their joint venture's growing own index.
Metasearch. Dogpile, MetaCrawler, WebCrawler, SearXNG and Presearch merge other engines' results by design.
Filtered wrappers. Kids' search engines are almost universally a Google or Bing instance with SafeSearch on and a block-list over the top.

So a plausible listicle of "50 search engines you can use instead of Google" may contain three or four distinct indexes. That is the single most useful thing to know before reading one.

Is the number going up or down?

Both, and the traffic runs in each direction, which makes the trend more interesting than the usual story of inevitable consolidation.

Losses since 2023 alone. Gigablast went offline without warning in early April 2023 after twenty-one years — one person, written in C++, no venture capital, an existence proof that a web-scale independent index could be built and run outside the duopoly. Neeva, a subscription engine with its own crawler founded by former Google executives, announced its shutdown around 20 May 2023 and closed its consumer search product on 2 June 2023. You.com's consumer search engine is gone, the company having pivoted to selling web-search APIs to AI developers. And Ask.com — thirty years old, a direct descendant of Ask Jeeves — closed on 1 May 2026, IAC stating that it had decided to discontinue its search business.

Gains over the same period. Brave completed its transition to full index independence in April 2023, the same month Gigablast disappeared. European Search Perspective started crawling and serving in August 2025, the first serious attempt at a new European general index in years. Perplexity moved from reselling to running its own crawl. Mojeek's index has grown steadily through the period.

The cost argument that supposedly makes new indexes impossible deserves examination too. Marginalia's operator has described running an index of a few hundred million documents on hardware costing roughly $200 a month. That does not make anyone a Google competitor and nobody claims it does — but it does mean the floor for a real, useful, independent index of a few hundred million pages is closer to a phone bill than to a data-centre budget. The barrier to a comprehensive index is enormous. The barrier to a good index of a chosen slice of the web is not.

There is also a new source of demand pushing in the opposite direction from consolidation. AI systems need web retrieval, cannot always get it from the two dominant engines on acceptable terms, and have become paying customers for independent indexes — Brave's search API and Mojeek's API both sell into that market, and You.com's entire surviving business is that market. For the first time in twenty years there is a commercial reason to build an index that is not "compete with Google for consumers."

What number should you actually use?

If you need a figure for a piece of writing or a citation, state what you counted. These are defensible, in descending order of usefulness:

"Around a dozen organisations worldwide operate their own web-scale crawler and index" — accurate, checkable, and the number that explains the shape of the market.
"Fewer than ten general web indexes matter outside China and Russia" — also accurate, and closer to what a Western reader experiences.
"Hundreds of consumer search engines are available" — true, but almost all of them are interfaces over the same handful of indexes, and the sentence is misleading without that qualification.

What you cannot honestly write is a single precise global total. Nobody is counting, the categories are not agreed, engines appear and vanish without announcements — Gigablast simply stopped responding one day and no statement was ever issued — and the largest measurement dataset in the field lumps every engine outside the top six into a single bucket. A reference that gives you a confident-sounding exact number for this question is telling you something about its methods rather than about search.

Frequently asked questions

How many search engines are there in the world?

There is no register and no census, so no exact figure exists. The useful count is of organisations that crawl and index the open web themselves: roughly a dozen worldwide, including Google, Microsoft Bing, Yandex, Baidu, Naver, Seznam, Sogou, Cốc Cốc, Brave, Mojeek, Marginalia and Perplexity. Hundreds of consumer search engines exist, but most are interfaces over those indexes.

How many search engines have their own index?

Very few. Building a web-scale index means continuously crawling, storing and ranking billions of documents. Counting generously, about a dozen organisations do it worldwide; counting strictly, requiring a global rather than national crawl, closer to six. Everything else — including many well-known privacy engines — buys results from Google or Microsoft, or blends bought results with a small index of its own.

Why do lists of search engines disagree so much?

Because they count different things and rarely say which. Some count brands, so Dogpile, MetaCrawler and WebCrawler appear as three engines when they are one company's properties on shared infrastructure. Some count every site with a search box. Some count software like SearXNG once, others count each public instance. None of these definitions is wrong; they are simply not comparable.

What share of search does Google have?

StatCounter Global Stats put Google at 91.31% worldwide across all devices in July 2026. That figure measures page views on the sites StatCounter tracks, not users or queries, so treat it as a proxy. Separately, the US district court that ruled on Google's search antitrust liability in August 2024 found roughly 90% of general search and about 95% on mobile.

Are new search engines still being built?

Yes, though slowly and against a strong current of closures. Brave completed its index independence in April 2023, European Search Perspective — the Qwant and Ecosia joint venture — began serving from its own index in August 2025, and Perplexity moved from reselling to crawling. Over the same period Gigablast, Neeva, You.com's consumer search and Ask.com all ended.

How much does it cost to run a search engine?

It depends entirely on ambition. A comprehensive global index costs a data centre. Marginalia's operator has described running an index of a few hundred million documents on hardware costing about $200 a month. The barrier to matching Google is enormous; the barrier to a genuinely useful index of a chosen slice of the web is much lower than commonly assumed.

How many search engines does this site cover?

Around 55 individual engines have their own profile here, spanning major global engines, privacy and AI engines, regional engines, metasearch, kids and safe search, and engines that no longer operate. That is a curated set rather than an attempt at an exhaustive list, because an exhaustive list of consumer search brands is neither achievable nor especially informative.

How many search engines are there?