Best Tor Search Engine: Top Picks for 2026

What a Tor Search Engine Can and Cannot Do

A Tor search engine indexes .onion sites — hidden services reachable only through the Tor Browser — but no single engine covers the entire dark web. When you open Tor Browser, the default search box uses DuckDuckGo, which indexes the surface web and does not list hidden services[4]. This surprises many first-time users: you can search for clearnet privacy guides, but pasting an .onion URL into DuckDuckGo returns nothing. That is expected behaviour, not a failure of Tor.

'Dark web search engine' usually means one of two things: an onion indexer (Ahmia, Torch) that crawls .onion domains, or a privacy-respecting clearnet engine (DuckDuckGo on Tor) that protects your query from third-party tracking. The two categories do not overlap. If you need to discover hidden services, you must use a dedicated onion indexer; if you want private access to Wikipedia or news sites, DuckDuckGo's onion service is already configured in your browser[4][2].

No onion indexer sees every hidden service. Site operators can block crawlers with robots.txt, and many do. Ahmia honours those instructions, so its index is smaller but more predictable. Torch ignores robots.txt and claims over 4 million pages, though that figure is self-reported and unverified[1]. Both engines will miss invite-only forums, private marketplaces, and services that change their .onion address frequently.

A conditional scenario: you search for 'secure messaging' in a regular search engine and paste the result — an .onion link — back into Google. You see zero results. This is normal. Clearnet search engines do not index .onion domains, because those addresses resolve only inside the Tor network. You must visit the link in Tor Browser and, if you want to find similar services later, use an onion indexer from the start.

Onion search engines cannot guarantee safety. Ahmia filters child abuse material before results become searchable and publishes a public MD5 blacklist for other indexes to reuse[3], but no engine removes scam sites, phishing clones, or malware. Every .onion link requires independent verification — check the address against community lists, compare results across two engines, and never enter credentials on a site you reached through search alone.

Ahmia: The Filtered Index That Prioritizes Legitimacy

Ahmia is the most widely recommended onion search engine among privacy projects and researchers because it applies content moderation before results become searchable. The engine filters child sexual abuse material and maintains a public blacklist of hashed onion addresses, published as MD5 sums so other indexes can reuse the list[3]. This filtering is why Ahmia's results feel sparse compared to unfiltered alternatives — you see fewer links, but each one has passed a documented abuse check.

Ahmia is operated by a Finnish nonprofit and appears in Tor Project documentation as a vetted resource. The engine honours robots.txt files, meaning a site owner can prevent Ahmia from indexing a page while the page still appears in Ahmia's list of known online services. This compliance shrinks the index further but keeps your search within a more predictable boundary: you will not accidentally land on a service whose operator explicitly blocked crawlers.

You can reach Ahmia through its official onion address or via a clearnet mirror. The clearnet version lets you preview results without opening Tor Browser, though visiting the actual .onion links requires Tor. Both versions query the same filtered index, so you see identical results either way.

Ahmia suits you if legitimacy matters more than exhaustive coverage. Use it when you want to explore hidden services without wading through scam clones and abuse material, and when you prefer an engine whose filtering policy is public and auditable[3]. The trade-off is a smaller result set — many hidden services block Ahmia's crawler, and others never appear because they fail the abuse filter. If you need the broadest possible crawl, Torch indexes more pages but applies no content moderation[1].

Before visiting any result, check the onion address against Ahmia's public MD5 blacklist or cross-reference it in a second engine. No search tool removes every malicious link, and Ahmia's filtering targets abuse material, not phishing or malware. Treat each .onion domain as untrusted until you verify it independently.

Torch: The Oldest Unfiltered Onion Index

Torch launched on October 10, 2013, making it one of the first dedicated .onion search engines[5]. The engine's self-reported index covers over 100,000 onion sites and more than 4 million pages, though independent verification of these figures does not exist[1]. You will see Torch described as indexing '1 billion pages' in older forum posts — that claim has no traceable source and likely confuses Torch with Haystak, whose operator reported 1.5 billion pages without third-party confirmation[1]. The documented size is large but far smaller than the billion-scale myth.

Torch applies no content filtering. You will encounter active marketplaces, research archives, political forums, and a high proportion of scam clones, dead links, and explicit material in the same result set. This lack of moderation is the central trade-off: broader coverage than Ahmia, but zero vetting. The engine does not honour robots.txt, so it indexes pages whose operators explicitly blocked other crawlers. If you need to cast the widest net and accept manual verification of every result, Torch delivers more links than any filtered alternative.

The engine displays heavy advertising, and uptime has been unreliable in documented periods[1]. You may find the onion address unreachable for hours or days, then working again without explanation. This instability makes Torch unsuitable as your only discovery tool — treat it as a supplementary index when Ahmia returns too few results.

Use Torch when you are researching the full scope of hidden services and have the time to cross-reference each link. Check every .onion address against Ahmia's public blacklist or a second engine before visiting, and never assume a result is legitimate because it appears high in the list[3]. Torch suits experienced users who understand that 'unfiltered' means you perform all safety checks yourself. For vetted browsing of legitimate services, start with Ahmia instead.

Haystak and DarkSearch: Aggregators with Different Trade-offs

Haystak claims to index 1.5 billion pages, making it the only onion search engine whose operator has publicly stated a billion-scale figure[1]. That number remains unverified by independent researchers, and you should treat it as a marketing claim rather than a documented fact. The engine uses an ad-supported model on its free tier and offers a premium subscription that removes advertisements and unlocks full search features. Haystak has experienced downtime and domain changes in documented periods, which means you may find the service unreachable when you need it most.

DarkSearch takes a simpler approach. The engine aggregates results from multiple onion indexes and applies some filtering, though the filtering policy is less transparent than Ahmia's public blacklist[3]. DarkSearch is run by a small team and offers a free API, which makes it useful for security researchers who want programmatic access to onion search results without scraping HTML. The interface is cleaner than Haystak's ad-heavy layout, and the engine has maintained more consistent uptime in community reports.

Both engines are secondary options. Use Haystak when Ahmia and Torch miss a specific resource and you want the largest self-reported index, accepting that you cannot verify the size claim and may encounter ads or downtime. Choose DarkSearch when you need API access for research scripts or prefer a simpler interface with aggregated results. Neither engine replaces Ahmia's documented filtering or Torch's unfiltered breadth[1] — they fill gaps when your first search returns nothing.

Before visiting any result from either engine, cross-reference the .onion address in Ahmia or check it against the public MD5 blacklist[3]. Haystak and DarkSearch do not publish abuse filtering as detailed as Ahmia's, so you perform the safety check yourself. If you are exploring hidden services for the first time, start with Ahmia's vetted index; return to these aggregators only when you need coverage beyond what filtered and unfiltered leaders provide.

DuckDuckGo on Tor: Why It's the Default but Not an Onion Search Engine

DuckDuckGo has been Tor Browser's default search engine since version 6.0.6, when it replaced Disconnect after that service lost access to Google's results and fell back to lower-quality Bing indexing[4]. This default setting confuses many first-time Tor users: you open the browser, type a query into the search box, and expect to discover hidden services. Instead, you see clearnet results — Wikipedia articles, news sites, privacy guides — but zero .onion links. DuckDuckGo does not index hidden services, and that is intentional design, not a limitation you can work around.

The engine's value on Tor is privacy, not onion discovery. DuckDuckGo does not store your search history, does not build a tracking profile, and does not personalise results based on past behaviour. When you use Tor Browser's 'Onionize' option, the browser automatically switches you to DuckDuckGo's onion service, keeping your query inside the Tor network from start to finish[4]. The onion version offers an '/html' endpoint that works with JavaScript completely disabled, a configuration most other search engines do not support[2]. This makes DuckDuckGo the safest way to search the surface web through Tor, but it does nothing to help you find .onion sites.

If you need to discover hidden services, you must use a separate onion indexer — Ahmia for filtered results, Torch for unfiltered breadth[1], or an aggregator like DarkSearch. DuckDuckGo and onion search engines serve different purposes, and you will use both depending on whether you are looking for clearnet privacy resources or exploring .onion domains. The default search box in Tor Browser handles the first task; for the second, bookmark an onion indexer and visit it directly.

For a broader introduction to vetted hidden services beyond search engines, see our checklist of the best dark web sites.

Excavator and Other Niche Onion Indexers Worth Knowing

Excavator is a specialised onion indexer that focuses on cryptocurrency services and marketplace listings rather than the full spectrum of hidden services. You will find vendor pages, exchange mirrors, and forum threads about digital currencies, but almost no academic archives, political forums, or privacy tools. This narrow scope makes Excavator useful when you are researching darknet markets or crypto-related services and want results filtered by topic rather than by content moderation. The engine does not apply abuse filtering like Ahmia, so you still perform all safety checks yourself.

Coverage is the central trade-off. Excavator indexes fewer total pages than Torch[1] and offers no filtering advantage over Ahmia, but it surfaces marketplace-specific pages that general engines may rank lower or miss entirely. Use Excavator as a supplementary tool when Ahmia and Torch return too many unrelated results for a market query. For a current overview of active marketplaces and their verified addresses, see our guide to dark web markets still active.

Other niche indexers exist for specific communities — forum archives, academic libraries, encrypted email providers — but most operate intermittently and lack public documentation. You will encounter these engines through community recommendations rather than search, and uptime is unpredictable. Treat any niche indexer as a secondary discovery method, not a replacement for the documented engines covered earlier.

The Hidden Wiki is not a search engine. It is a manually curated link directory, and many versions circulate with identical names but different operators. Scam risk is high: cloned directories add phishing links to legitimate categories, and you cannot verify which version is 'official' because no single entity controls the name. If you find an .onion address through any directory, verify it against at least one independent source — cross-reference the link in Ahmia, check community forums, or compare it to addresses listed in vetted guides. Never assume a directory entry is safe because it appears under a trusted category heading.

How to Choose: Matching the Engine to Your Task

You choose an onion search engine by matching its filtering policy, index size, and uptime to your specific need. Use Ahmia when you want vetted results with documented abuse filtering and can accept a smaller index. Choose Torch when you need the broadest possible crawl and are prepared to verify every link manually[1]. Select DuckDuckGo for privacy-respecting clearnet search inside Tor Browser, not for discovering .onion sites[4]. Turn to Haystak or DarkSearch when you need bulk coverage or API access for research scripts[1].

The table below summarises the core trade-offs:

Before visiting any result, verify the onion address. Cross-reference it in Ahmia's public MD5 blacklist[3], check it in a second engine, or compare it to addresses in vetted community guides. No search tool removes every malicious link — Ahmia filters abuse material, not phishing or malware.

Expect dead links. Hidden services go offline without notice, and indexes do not update in real time. If a result returns a connection error, try a second address from the same query or switch to another engine.

Never enter credentials on a site you discovered through search alone. Verify the service through an independent channel — a trusted forum thread, a security researcher's published list, or cross-reference in multiple engines. Treat every .onion domain as untrusted until you confirm it independently.

When you need to explore beyond individual searches, consult our checklist of the best dark web sites for vetted starting points across categories.

Common Mistakes When Searching the Dark Web

First-time Tor users make five recurring mistakes that expose them to phishing, malware, or wasted time. Each mistake has a documented consequence and a straightforward correction.

Pasting .onion URLs into Google or clearnet search engines. Onion addresses work only inside Tor Browser; a clearnet search engine cannot resolve them and will return no results or, worse, phishing sites that mimic the service you wanted. The Tor Project's support documentation repeatedly warns that .onion domains are network-specific[4]. The correct action: open Tor Browser, then paste the address directly into the browser's address bar or use an onion indexer like Ahmia to discover services.

Trusting the first result from an unfiltered engine without verification. Torch indexes over 100,000 sites with no content moderation[1], meaning the top result for a marketplace or forum query may be a scam clone. You will lose time or credentials if you assume search rank equals legitimacy. Always cross-reference the .onion address in Ahmia's public blacklist[3], check a second engine, or compare it to addresses in vetted community guides before visiting.

Clicking links without checking the onion address format. Legitimate v3 onion addresses are 56 characters long and contain only lowercase letters and digits. Shorter addresses or those with uppercase letters, special characters, or unusual extensions are either v2 addresses (deprecated and insecure) or phishing attempts. Verify the format before you click; if the address looks wrong, search for the service name in Ahmia to find the correct domain.

Assuming a search engine's existence means indexed sites are safe. Ahmia applies documented filtering to remove child abuse material, but no engine removes phishing mirrors, malware hosts, or scam marketplaces. Haystak and DarkSearch do not publish filtering policies as detailed as Ahmia's[3], and Torch applies no filtering at all[1]. You perform the safety check yourself: verify addresses independently, never enter credentials on a site discovered through search alone, and treat every result as untrusted until confirmed.

Confusing the deep web with the dark web. The deep web is any content not indexed by clearnet search engines — password-protected email, subscription databases, private social media posts. The dark web is the subset of the deep web accessible only through anonymity networks like Tor[6]. When you search for 'deep web search engines', you often find guides to onion indexers, but those tools do not help you access your university library or a paywalled news archive. Use DuckDuckGo's onion service[4][2] for privacy-respecting clearnet search inside Tor; use Ahmia or Torch[1] when you specifically need to discover .onion hidden services.

The Tor Project's security guides and support threads document these patterns in user questions and incident reports[4]. You avoid all five mistakes by verifying every address, understanding what each engine indexes, and never assuming search results equal safety.

Verifying Search Results Without Putting Yourself at Risk

Every .onion address you find through search requires independent verification before you visit. Search engines surface both legitimate services and phishing clones, and no index removes every malicious link — Ahmia filters child abuse material, not scams or malware. You confirm an address is real by cross-referencing it in at least two independent sources: check Ahmia's public MD5 blacklist[3], compare the result in a second engine, or look for the address on the service operator's clearnet site or in a vetted community guide like our checklist of the best dark web sites.

Legitimate v3 onion addresses are always 56 characters long and contain only lowercase letters and digits. Any shorter address is either an outdated v2 domain (no longer secure) or a fake. If the format looks wrong, do not click — search for the service name in Ahmia to find the correct domain. Prefer addresses published with PGP-signed announcements from the site operator, though most small services do not maintain signing keys. When no signed announcement exists, multiple independent directory listings reduce (but do not eliminate) phishing risk.

Using Tor and searching the dark web is legal in the United States[7], and government agencies rely on Tor for secure communication — the CIA operates an onion service[7]. Running a Tor relay, including an exit relay, is also legal under U.S. law[8], though law enforcement has occasionally mistaken relay operators for criminals[8]. However, accessing illegal content remains a crime regardless of the browser you use. Search engines may surface such content even when you are not seeking it, so you must verify every result and avoid sites that clearly host prohibited material. Your internet service provider can see that you are using Tor (but not the content of your browsing)[6], and using Tor with a VPN can add a layer of privacy[6].

Never enter credentials on a site discovered through search alone. Treat every .onion domain as untrusted until you confirm it through an independent channel.

Who Each Engine Suits

If you’re using Tor Browser for the first time — choose Ahmia

Ahmia applies a public blacklist and honors robots.txt, so results are filtered before you see them. The trade-off is a smaller index, but that’s the point: you get vetted starting points rather than a wall of unverified links. First-time users avoid the most common mistake — clicking a scam clone on page one — because Ahmia removes abuse material and many junk entries before they reach you.

If you need maximum coverage for research — choose Torch

Torch indexes over 100,000 sites and 4 million+ pages, with no content moderation at all[1]. That breadth helps when you’re mapping a niche, looking for obscure archives, or documenting what exists rather than what’s safe. You must verify every link manually, because the absence of filtering means scam clones and phishing mirrors appear alongside legitimate services[1].

If you want private clearnet search inside Tor — choose DuckDuckGo

DuckDuckGo’s onion service lets you search the surface web without leaving the Tor network[4]. It does not index .onion domains, so it won’t help you discover hidden services. Use it when you’re researching a topic, reading news, or checking a clearnet source while keeping your browsing private from the search provider[4][2].

If you’re building a research script or need bulk data — choose Haystak or DarkSearch

Haystak claims 1.5 billion indexed pages, though the figure is unverified, and the free tier shows ads[1]. DarkSearch aggregates multiple indexes and offers an API, which helps when you need programmatic queries rather than manual browsing. Both have less transparent filtering policies than Ahmia, so treat their results as raw material for verification, not as a safety filter[3][1].

If you only need a curated starting point — skip engines and use a vetted directory

The Hidden Wiki is not a search engine; it’s a manually curated link directory with cloned versions that add phishing links to legitimate categories. You cannot verify which version is ‘official’ because no single entity controls the name. If you find an .onion address through any directory, cross-reference it in Ahmia or another independent source before visiting — a directory entry under a trusted heading does not make the link safe.

Final Verdict: What to Choose

For most readers, we recommend Ahmia as the default onion search engine. It applies a public blacklist and honors robots.txt, so you see fewer scam clones before you click. The smaller index is a feature, not a flaw, when you are new to Tor.

Choose Torch only when you need maximum coverage for research and are prepared to verify every result manually, because it applies no content moderation[1]. Choose DuckDuckGo’s onion service for private clearnet search, not for discovering .onion sites[4]. Avoid unfiltered engines as your first stop and never enter credentials on a site found through search alone.

Start with our checklist of the best dark web sites, then cross-reference any address in Ahmia’s public MD5 blacklist[3] before visiting.