Ahmia: The Tor Onion Search Engine

By Bill Hartzer · Updated 27 October 2026

What Ahmia is, and why it is not a web index

Ahmia is a search engine for onion services — websites that can be reached only through the Tor network, a system that routes traffic through several volunteer-run relays so that neither the visitor nor the site needs to reveal its location. Onion services are what people usually mean by "the dark web". Ordinary search engines do not index them, because their crawlers cannot reach them.

Ahmia runs its own crawler and its own index. It does not re-serve another engine's results. But what it indexes is a separate network rather than the public web, which is why this site classes it as a specialist index rather than an own-index web engine. Its stated aim is to be "the premier search engine for services residing on the Tor anonymity network."

Two facts set Ahmia apart from other onion search engines. The first is its relationship with the Tor Project: Tor developers mentored its development through Google Summer of Code in 2014, and the Tor Project profiled it in its "Tor at the Heart" series in December 2016. The second is an explicit, published abuse-content filter. That filter is the reason a reference site can describe Ahmia plainly, and it is explained in full below.

Ahmia is reachable both on the ordinary web, at ahmia.fi, and as an onion service. This page gives only the ordinary web address and does not print any onion address.

How Ahmia finds onion services

Discovery is the hard problem in onion search, and it is harder than it used to be. Onion services are not linked from the ordinary web in any systematic way. Since version 3 onion services were introduced in 2017 — and version 2 was retired in 2021 — their addresses can no longer be collected from Tor's own directory system. A search engine for onion services therefore has to find addresses the slow way.

Ahmia does it in two ways. It crawls both the ordinary web and onion space for links that point to onion services, an approach the Tor Project's 2016 profile described as requiring "heavy" infrastructure across both spaces. And operators can submit their own addresses through an "Add service" page. Historically, Wikipedia records discovery inputs from GlobaLeaks submissions and Tor2web statistics.

The software is open source end to end. There are three components: ahmia-crawler, built on the Scrapy crawling framework; ahmia-index, built on Elasticsearch, the search and indexing engine; and ahmia-site, the Django web application users see. Anyone can read how the engine is built, which is unusual for a search engine of any kind.

A secondary source — a comment on the Tor Project's 2016 post — says Ahmia respects robots.txt, the file a site uses to tell crawlers what not to fetch. That has not been confirmed against Ahmia's own documentation for this page.

On size, Ahmia publishes no current count. The only dated figure found is about 5,000 sites indexed as of July 2015. For a sense of the network's scale, Tor Metrics estimated an average of about 656,000 unique version 3 onion addresses per day over 26 September to 25 October 2026. That is an extrapolated estimate rather than a count, and the 30-day mean was computed from Tor's daily data by this site's research. The two numbers are not comparable: most onion addresses host no public website at all, and many exist only briefly.

What Ahmia filters, and why

Ahmia filters content in three layers, and it says so publicly.

A blocklist of child sexual abuse material (CSAM). Onion services known to host such material are removed from the index, and Ahmia publishes a blacklist of banned services.
User reporting. Visitors can report abusive material for removal.
Search filtering, since late 2023. Ahmia filters all sexual and suspicious searches, not only results. The 2024 research paper co-authored by Ahmia's founder dates this decision to November 2023; Wikipedia dates it to October 2023. Both dates are printed here because the sources disagree.

The reason for the 2023 expansion is documented in unusual detail. A study published in Scientific Reports on 3 April 2024 — by Juha Nurmi and colleagues from Tampere University, the child-protection organization Protect Children and other institutions — analyzed Ahmia's own search logs. In the authors' words: "We analyse 110,133,715 search sessions from the Ahmia.fi search engine and discover that 11.1% seek CSAM." The same study examined 176,683 onion domains from 2018 to 2023 and found that about one-fifth shared such material; in a January 2023 sample, 221 of 1,000 domain addresses did.

Ahmia's response was to stop serving those searches and to redirect them toward help resources. The paper frames this as a public-health intervention — an attempt to reach people at the moment of searching, rather than only to remove material after the fact.

One caution belongs with these figures, and it matters. The data come from one search engine's logs. They describe what people searched for on Ahmia, not what all Tor users do, and they should not be quoted as a measure of the Tor network as a whole. What they do establish is that the filter responds to a measured problem on this engine, not to a hypothetical one.

Ahmia has also acted against fraud. In July 2015 it published a list of hundreds of cloned onion sites used to scam visitors — for example by swapping in a different Bitcoin address on a copied page. Its live site still warns of a fake, man-in-the-middle clone of Ahmia itself.

Who runs it, and how it is funded

Ahmia is led by Juha Nurmi, described on the site as "the founder and project leader of the Ahmia search engine project." Nurmi is a Finnish security researcher; at the time of the 2024 study he was a doctoral researcher in Tampere University's NISEC group. Wikipedia also records an affiliation with the Hermes Center for Transparency and Digital Rights, an Italian organization. No company behind Ahmia, and no formal legal entity, could be established for this page.

There is no commercial business model. No advertising was found on the clearnet site, and there is no subscription. The Tor Project's 2016 profile called Ahmia a "low-budget search engine" that was always looking for volunteers and funding. Ahmia's about page says Nurmi has worked on projects funded by commercial and government sources, without naming who funds Ahmia itself. Funding sources are therefore unconfirmed.

That structure has consequences. A volunteer project with no revenue is not exposed to advertising pressure, and its code is open for inspection. It also has no paid staff to guarantee uptime, crawl coverage or response times, and its continued operation depends on the people running it.

History

The ahmia.fi service was already running in April 2014. On 22 April 2014 Nurmi posted to the Tor Project's developer mailing list as "a student who is starting to work with ahmia.fi search engine as a part of Google Summer of Code," with Tor developers as mentors. Wikipedia dates the launch to 2014. The service predates that summer by an unknown period, so the accurate statement is that Ahmia was developed with the Tor Project through Google Summer of Code 2014; its exact first launch date is not established.

By July 2015 it had indexed about 5,000 sites and published its list of fraudulent clones. On 5 December 2016 the Tor Project profiled it in "Tor at the Heart." The retirement of version 2 onion addresses in 2021 changed the discovery problem for every onion search engine. In late 2023 Ahmia expanded its filtering, and on 3 April 2024 the Scientific Reports study based on its logs was published. As of 27 October 2026 the clearnet site loads, with search (JavaScript required), an "Add service" page, and its blacklist and policy pages.

Who it suits, and who it does not

It suits researchers, journalists and students who need to understand what is published on onion services, and who want a search engine whose code, filtering policy and research record are public. Its association with the Tor Project, its open-source stack and its documented filtering make it the onion search engine most suitable for citation in serious work. Its search logs underpin the largest published study of what people search for on Tor.

It is a poor choice for anyone expecting the completeness of a web search engine. The only published index figure is a decade old. Discovery depends on links and submissions, so services that are never linked or submitted will not appear. Onion services come and go quickly: the Dizzy study, a large academic crawl run in 2021 and 2022 and published in 2023, found only 52.2% of the onion domains it reached available on a typical day. That crawl saw only what its own link discovery could reach, but the direction is clear — any onion index is always partly stale. A volunteer project also offers no service guarantees.

It is also deliberately the wrong tool for anyone seeking the material it filters. That is the design, and it is the right design. The 2024 Scientific Reports paper sets out the evidence behind it.

Common misconceptions

"Ahmia is the dark web's Google." It is not, in scale or in method. Google's index is built from a public web where pages link to one another freely. Ahmia's is built from a network designed to resist discovery, and the only published index figure is about 5,000 sites in 2015 — set against Tor Metrics' estimate of roughly 656,000 onion addresses a day in late 2026, most of which host no public site.

"An onion search engine shows everything on the dark web." No onion search engine can. Since version 2 addresses were retired in 2021, there is no directory to harvest; an address that nobody links to and nobody submits is effectively invisible.

"Filtering makes the research data meaningless." The 2024 study's figures came from Ahmia's search logs, and they are what prompted the expanded filter. They describe Ahmia's users rather than all Tor users, and that limitation should travel with every quotation of them.

"Using Ahmia's clearnet site is anonymous." Ahmia.fi is an ordinary website. Reaching it through a normal browser does not route anything through Tor, and Ahmia's own warning about a fake clone of itself is a reminder that a familiar name on a page is not proof of who is running it. The Ahmia about page describes the project and its filtering in its own words.

Frequently asked questions

What is Ahmia?

Ahmia is a search engine for Tor onion services — websites reachable only through the Tor anonymity network. It runs its own open-source crawler and index, was developed with Tor Project mentors through Google Summer of Code 2014, and is led by Finnish security researcher Juha Nurmi. It is reachable on the ordinary web at ahmia.fi and was operating as of October 2026.

What does Ahmia filter?

Ahmia removes child sexual abuse material from its index through a published blocklist, accepts user reports of abusive content, and since late 2023 filters all sexual and suspicious searches. A 2024 Scientific Reports study of 110,133,715 Ahmia search sessions found that 11.1% sought such material, which prompted the expansion. Ahmia redirects those searches toward help resources rather than serving results.

Is Ahmia run by the Tor Project?

No. Ahmia is an independent project led by Juha Nurmi, its founder. The Tor Project's developers mentored its development through Google Summer of Code in 2014, and the Tor Project profiled it in its "Tor at the Heart" series in December 2016, but Ahmia is not a Tor Project product. No company or formal legal entity behind Ahmia could be established.

How many sites does Ahmia index?

Ahmia publishes no current count. The only dated figure found is about 5,000 sites in July 2015. For comparison, Tor Metrics estimated about 656,000 unique version 3 onion addresses per day on average from 26 September to 25 October 2026, but that is an extrapolated estimate, and most onion addresses host no public website, so the two figures cannot be compared directly.

When did Ahmia start filtering sexual searches?

In late 2023, though the sources disagree on the month. The 2024 Scientific Reports paper co-authored by Ahmia's founder says Ahmia decided in November 2023 to filter all sexual and suspicious searches; Wikipedia dates the change to October 2023. Before that, Ahmia already maintained a blocklist of services hosting child sexual abuse material.

How does Ahmia make money?

It does not, in any commercial sense. No advertising was found on its clearnet site and there is no subscription. The Tor Project's 2016 profile described Ahmia as a low-budget project always looking for volunteers and funding. Its about page does not name who funds it, so its funding sources remain unconfirmed.

Can Ahmia find every onion site?

No, and no onion search engine can. Since Tor retired version 2 onion addresses in 2021, addresses can no longer be collected from Tor's directory system, so Ahmia relies on finding links and on operators submitting their own services. A service nobody links to or submits will not appear, and many onion services are offline on any given day.

Ahmia