A directory and guide for acces…

Dark Web Search Engine Links 2026

Ahmia leads with 18,000+ indexed onion addresses, while most dark web search engines cover under 5% of Tor services and 83% of links are dead.

Published: Updated: November 20, 202614 minute read

Ahmia is the leading dark web search engine, indexing 18,069 onion addresses in a single day and ranging between 18,000–22,000 addresses weekly—outperforming five other engines combined (1). Ahmia is accessible via clearnet (ahmia.fi) and .onion addresses; verify the current .onion link through Ahmia's clearnet site before use, as onion addresses change frequently.

Torch collected only 4,190 addresses in the same March-April 2025 study, ranking last among six tested engines (2). Combined, six major search engines indexed just 36,028 unique addresses after deduplication—representing only 4-5% of the estimated 800,000 onion services on Tor (2).

Expect most links to be dead: research found 83% of .onion addresses from search engines were unreachable (3). The Tor ecosystem is highly unstable, with many hidden services disappearing shortly after emergence (4).

Ahmia – Leading Index with 18,000+ Onion Addresses

Ahmia indexed 18,069 onion addresses in a single day during March-April 2025 research, with weekly totals ranging between 18,000 and 22,000 addresses (1). This performance outpaced five other dark web search engines combined in the same study. Ahmia's clearnet site is accessible at ahmia.fi, where you can verify the current .onion address before accessing the hidden service version.

Ahmia maintains an intentional exclusion blacklist containing over 46,000 hashed addresses to filter illegal content (1). Research found that 11,809 out of 48,745 unique v3 addresses discovered through crawling appeared on this blacklist. This filtering explains why Ahmia's results exclude certain categories of hidden services, prioritizing legal and ethical content over comprehensive coverage.

The engine's dual-access model lets you search from either clearnet or Tor. Visiting ahmia.fi through a standard browser provides search functionality without requiring Tor Browser, though results link to .onion addresses you'll need Tor to access. The .onion version offers the same search interface with full anonymity for users already connected to the Tor network.

Ahmia's index size makes it the most comprehensive publicly available dark web search engine based on documented research. The 18,000+ address range represents roughly 3% of the estimated 650,000 onion services observed on Tor (5), but this is significantly larger than any competing search engine's verified coverage. For users seeking the widest selection of discoverable hidden services, Ahmia provides the largest starting point.

The engine updates its index regularly through automated crawling. Unlike manually curated directories that rely on user submissions, Ahmia discovers addresses by following links from known onion sites. This approach captures newly launched services faster than manual methods, though it also means the index includes many short-lived or abandoned addresses.

Torch and Other Search Engines – Limited Coverage Compared to Ahmia

Torch collected only 4,190 onion addresses in the March-April 2025 research study, ranking last among six tested search engines (2). This represents less than one-quarter of Ahmia's single-day collection and demonstrates the significant performance gap between leading and secondary dark web search engines.

Six major onion search engines and repositories combined collected only 36,028 unique addresses after deduplication in 2025 (2). This total represents approximately 4 to 5% of the Tor Project's estimated 800,000 onion services at that time. Even pooling results from multiple engines leaves the vast majority of the dark web unindexed and undiscoverable through search.

The research tested Ahmia, Torch, and four other engines whose individual performance fell between these two extremes. No single engine beyond Ahmia approached five-figure address counts, and the combined total of all six remained below 40,000 unique addresses. This clustering shows that dark web search technology faces fundamental limitations that affect all engines, not just smaller or less sophisticated ones.

Torch's lower performance doesn't necessarily indicate technical inferiority. The engine may apply stricter filtering, focus on specific content categories, or prioritize address quality over quantity. Without detailed methodology from Torch's operators, the address count alone doesn't explain why coverage differs so dramatically from Ahmia's. Users should recognize that Torch will return fewer results for most queries, but those results may still include relevant hidden services not found through Ahmia.

Other engines in the study remain unnamed in available research summaries, making direct comparison impossible. If you're evaluating a dark web search engine not mentioned here, assume its coverage falls somewhere in the 4,000–18,000 address range unless you find documented evidence otherwise. Engines claiming to index hundreds of thousands of addresses without published verification likely overstate their actual coverage.

The 4–5% indexing rate means that searching multiple engines increases your chances of finding specific hidden services. A site missing from Ahmia's index might appear in Torch or another engine, though the probability remains low. For comprehensive searches, query at least two engines and cross-reference results with community directories and forums where users share verified links.

Research found that 83% of .onion addresses collected from search engines and OSINT sources were unreachable, with only 1,157 out of approximately 6,800 seeds (17%) accessible for crawling (3). This failure rate applies to addresses specifically gathered for research purposes, suggesting that random links from search results face even higher odds of being dead.

The Tor ecosystem is heavily populated by replicas of .onion sites and is highly unstable, with a large proportion of hidden services disappearing shortly after they emerge (4). A 2026 survey confirmed this instability, documenting that many addresses observed by Tor directory caches never respond to connection attempts or vanish within days of first appearance.

Hidden services disappear for multiple reasons. Operators shut down sites intentionally, servers go offline due to technical failures, law enforcement seizes infrastructure, or services migrate to new addresses without leaving redirects. Unlike clearnet domains with DNS records and hosting contracts, .onion addresses can vanish instantly with no trace or explanation.

The largest research collection assembled 482,614 unique v3 onion addresses by combining six search engines, clearnet search engines, DNS leakage data, and 20 days of crawling 2.9 million pages (1). This represented the most comprehensive onion address dataset known at the time. Even this extensive collection accounted for only 25% of observed blinded public keys, meaning three-quarters of Tor hidden services remained undiscovered (1).

The 482,614 addresses were responsible for 66% of all successful service descriptor downloads despite representing just 25% of observed services (1). This heavy-tailed distribution confirms that most onion addresses see little to no traffic, while a small fraction of services handle the majority of Tor hidden service activity. The rarely-accessed addresses are more likely to be abandoned, spam, or test sites that operators never maintain.

Link decay accelerates over time. An address that worked when added to a search engine's index may fail weeks or months later when you attempt to visit. Search engines don't continuously verify every indexed address, so their databases accumulate dead links faster than crawlers discover new ones. Expect that any .onion link more than a few months old has a high probability of failure.

For users, this means you'll encounter "Unable to connect" errors frequently when clicking search results. This is normal behavior on the dark web, not a problem with your Tor Browser or connection. Try multiple results for your query, and if a specific service is critical, search for recent community discussions where users confirm working addresses.

The Scale of the Dark Web: 650,000+ Onion Services

Tor directory caches observed approximately 590,000 to 720,000 v3 onion services as of October 2024, with a mean around 650,000 (5). This count comes from Tor Project infrastructure that tracks service descriptor publications, providing the most authoritative estimate of dark web size available. The actual number of real sites versus generated spam remains uncertain.

The 650,000 figure represents unique .onion addresses that published service descriptors, not necessarily active or accessible sites. Many of these addresses may be spam, test services, or abandoned sites that still publish descriptors but don't respond to connections. The Tor Project's count includes all addresses that interact with the network, regardless of whether they host legitimate content or functional services.

Six major search engines combined captured just 36,028 unique addresses after deduplication, representing 4–5% of the estimated total (2). This gap between observed services and indexed addresses shows why comprehensive dark web search remains impossible. Even the most aggressive crawling and collection methods documented in research reached only 482,614 addresses (1), or roughly 75% of the observed total at that time.

The indexing gap exists because most hidden services don't link to each other. Crawlers discover new addresses by following links from known sites, but isolated services with no inbound links remain invisible. Operators may intentionally avoid linking to keep services private, or sites may exist in disconnected clusters that search engines never reach.

Spam and automatically generated addresses inflate the total count. Researchers and malicious actors can create thousands of .onion addresses programmatically, publishing service descriptors without hosting real content. These addresses appear in Tor's statistics but offer nothing to users. Distinguishing between legitimate sites and spam requires manual verification, which is impractical at scale.

The dark web's size fluctuates significantly. The 590,000–720,000 range observed in October 2024 (5) shows that tens of thousands of addresses appear and disappear within short periods. This volatility reflects the unstable nature of hidden services documented in the 2026 survey (4), where many addresses emerge briefly before vanishing.

For practical purposes, the searchable dark web is much smaller than 650,000 addresses. Focus on the 4–5% indexed by search engines, plus addresses shared in trusted directories and community forums. The remaining 95% consists of private services, spam, dead links, and isolated sites you'll never discover through conventional search methods.

Comparison Table: Dark Web Search Engine Coverage

Engine Addresses Indexed Research Period Filtering Policy Status
Ahmia 18,069 (single day); 18,000–22,000 (weekly) (1) March-April 2025 Blacklist of 46,000+ hashed addresses blocks illegal content (1) Leading engine; clearnet access at ahmia.fi
Torch 4,190 (2) March-April 2025 Unknown Last among six tested engines
Other Four Engines Combined with above: 36,028 total unique (2) 2025 Varies by engine Individual counts not published
All Six Combined 36,028 unique after deduplication (2) 2025 — Represents 4–5% of estimated 800,000 Tor services

This table shows documented performance from the same research period, making direct comparison valid. Ahmia's lead is substantial: its single-day count exceeds Torch's total by more than 4× and approaches the combined unique total of all six engines.

Checklist: Verifying an Unfamiliar Dark Web Search Engine

Before trusting a new dark web search engine, verify its legitimacy and safety through these steps:

  1. Check the .onion address format. V3 addresses are exactly 56 characters long before the .onion suffix. V2 addresses (16 characters) are no longer supported. Reject any address that doesn't match v3 format.

  2. Cross-reference the address in multiple sources. Find the .onion link mentioned in at least three independent locations: forums, community wikis, and trusted clearnet sites. Single-source addresses may be phishing attempts.

  3. Search for recent user reports. Look for discussions from the past 30 days where users confirm the engine works. Avoid addresses with no recent mentions or only old references.

  4. Visit from Tor Browser's "Safest" security level first. Disable JavaScript and other potentially exploited features before your initial visit. This limits damage if the site is malicious.

  5. Verify the site doesn't request personal information. Legitimate search engines don't require registration, email addresses, or cryptocurrency payments to perform searches. Reject any engine asking for credentials.

  6. Test with a generic query. Search for a common term and check if results link to known legitimate .onion addresses. If results are all dead links or redirect to suspicious sites, the engine may be compromised.

  7. Compare results against Ahmia. Run the same query on Ahmia and the unfamiliar engine. Legitimate engines should return some overlapping results for common searches.

  8. Check for clearnet presence. Established engines often maintain clearnet sites with documentation and verified .onion links. Absence of clearnet presence doesn't disqualify an engine, but it removes one verification method.

This checklist reduces risk when exploring new search engines, though no verification process eliminates all danger. Treat every unfamiliar .onion address as potentially hostile until you've confirmed its legitimacy through multiple independent sources.

How Many Onion Addresses Do Dark Web Search Engines Actually Index?

Ahmia indexes 18,000 to 22,000 onion addresses based on March-April 2025 research (1). This represents the largest documented index among publicly available dark web search engines. Torch indexes approximately 4,190 addresses (2), and four other tested engines fall between these extremes. Combined, six major engines collected 36,028 unique addresses after removing duplicates (2).

These numbers represent a small fraction of the dark web. With approximately 650,000 v3 onion services observed by Tor directory caches (5), even Ahmia's 22,000-address index covers only 3.4% of the total. The combined six-engine total of 36,028 addresses reaches just 5.5% coverage. No search engine indexes more than a small minority of hidden services.

Index size fluctuates as engines discover new addresses and remove dead links. Ahmia's weekly range of 18,000–22,000 (1) shows that thousands of addresses enter and exit the index regularly. This volatility reflects the unstable Tor ecosystem where hidden services appear and disappear constantly (4).

Larger indexes don't guarantee better search results. An engine that indexes 20,000 addresses but includes 15,000 dead links provides worse user experience than an engine with 5,000 addresses where 80% work. Ahmia's blacklist of 46,000+ addresses (1) shows that filtering reduces index size but improves result quality by removing illegal and harmful content.

For users, these numbers mean you should search multiple engines for comprehensive coverage. Ahmia provides the widest starting point, but addresses indexed by Torch or other engines may not appear in Ahmia's results. Expect that most hidden services you're looking for won't appear in any search engine, requiring alternative discovery methods like community directories and forum recommendations.

TorDex is a dark web search engine and directory that indexes .onion addresses, but it was not included in the March-April 2025 research study that tested Ahmia, Torch, and four other engines (1) (2). Without verified performance data, we can't confirm how many addresses TorDex indexes or how its coverage compares to Ahmia's 18,000+ addresses.

TorDex functions as both a search engine and a manually curated directory. Users can submit onion addresses for inclusion, and the site organizes links by category. This hybrid approach differs from pure crawlers like Ahmia, which automatically discover addresses by following links. Manual curation can improve quality by filtering spam, but it limits coverage since TorDex relies on user submissions and manual review.

The absence of TorDex from recent research suggests either smaller scale or less visibility in the academic community studying dark web infrastructure. If TorDex maintained an index comparable to Ahmia's, researchers likely would have included it in comparative studies. This doesn't mean TorDex is useless, but it's not a proven leader in dark web search based on available evidence.

If you're exploring TorDex, verify its current .onion address through trusted community sources before visiting. Like all dark web search engines, TorDex addresses can change, and phishing sites may impersonate legitimate engines. Cross-reference the address in multiple forums and check for recent user confirmations that the link works.

TorDex may serve as a supplementary tool alongside Ahmia and Torch. Its category-based directory structure can help users browse by topic rather than searching for specific terms. For comprehensive dark web searches, query multiple engines including TorDex if you have a verified working address, but prioritize Ahmia for the largest documented index.

What is the best dark web search engine in 2026?
Ahmia is the leading dark web search engine, indexing 18,069 onion addresses in a single day and ranging between 18,000–22,000 addresses weekly in March-April 2025 research, outperforming five other engines combined (1). Access Ahmia through its clearnet site at ahmia.fi to verify the current .onion address.

How many onion addresses do dark web search engines actually index?
Ahmia indexes 18,000–22,000 addresses (1), Torch indexes approximately 4,190 (2), and six major engines combined collected 36,028 unique addresses after deduplication (2). This represents only 4–5% of the estimated 800,000 onion services on Tor.

Why are so many dark web links dead or not working?
Research found that 83% of .onion addresses from search engines and OSINT sources were unreachable, with only 17% (1,157 out of approximately 6,800) accessible for crawling (3). The Tor ecosystem is highly unstable, with many hidden services disappearing shortly after they emerge (4).

How do I verify a dark web search engine link is safe?
Cross-reference the .onion address in at least three independent sources, verify it matches the 56-character v3 format, check for recent user confirmations within the past 30 days, and visit first with Tor Browser's security level set to "Safest." Use the verification chec

References

[1] Evaluating Onion Address Collection Methods — circumvention-corpus

[2] Dark Web Search Engines: How Torch and Ahmia Work

[3] Snorkeling in dark waters: A longitudinal surface exploration of unique Tor Hidden Services (Extended Version)

[4] Dark web survey reveals Tor is smaller, shakier and more duplicated than expected

[5] How Big Is the Dark Web? What Metrics Show | Dark Web Insight

Q&A

What is the best dark web search engine in 2026?

Ahmia is the leading dark web search engine, indexing 18,069 onion addresses in a single day and ranging between 18,000–22,000 addresses weekly in March-April 2025 research, outperforming five other engines combined (1). Torch collected only 4,190 addresses in the same study, ranking last among six tested engines (2).

Why are so many dark web links dead or not working?

Research found that 83% of .onion addresses from search engines and OSINT sources were unreachable, with only 17% (1,157 out of approximately 6,800) accessible for crawling (3). The Tor ecosystem is highly unstable, with a large proportion of hidden services disappearing shortly after they emerge (4).

Where to look next

These services are useful starting points for further research. Tools & services.