← Back to Directory
Logo: Mojeek

Mojeek

UK search engine that crawls and ranks the web with its own MojeekBot index, plus a no-tracking default for people and a Web Search API for products.

searchprivacyindependent-index
Visit WebsiteAsk an EU migration specialistFeedback & corrections

Mojeek is a United Kingdom search engine operated by Mojeek Limited. It answers web queries from an index it crawls itself with MojeekBot, then ranks with its own algorithms, rather than syndicating Google or Bing.

The product exists so there is a working alternative to Big Tech search that is not a privacy proxy in front of someone else's index. Mojeek's stated differentiator is that combination: a crawler-based web index plus a no-tracking privacy policy. By default it does not set cookies. The privacy policy says request logs keep time, path, optional referrer, and a separate browser log, with the IP address replaced by a two-letter country code.

Consumer search is available on the website, via official mobile apps, and as a browser search engine. Organisations can query the same web index through a JSON or XML Web Search API, use Focus to restrict queries to a defined set of sites without creating an account, or use Organisation Search to manage how their own domains appear.

Independent web indexOwn crawler (MojeekBot)No-tracking policy (claimed)UK-operatedWeb Search API

Shortlist when you need web results from a UK-operated independent crawler and index, plus a documented Search API, and you can live with thinner coverage than Google or Bing. Skip when you need image-first search, a published DPA/ISO pack, or a guarantee that optional summaries never leave UK systems. Consider Qwant if you need an EU (France) legal seat, or stay on Google/Bing if coverage is the hard requirement.

Key capabilities

Web results are produced from Mojeek's own crawler (MojeekBot) and ranker, not from Google or Bing syndication. The bot honours robots.txt and noindex/nofollow/nocache meta tags, rate-limits to at most one request per site per second, and publishes a verification method plus a public IP list at /mojeekbot.json. Useful when the evaluation criterion is index independence. Constraint: coverage and freshness will not match the US incumbents.

The privacy policy says Mojeek does not track users, does not set cookies unless the user agrees, and replaces IP addresses in logs with a two-letter country code. Preferences can be stored in an optional cookie or encoded in a shareable URL so there is no account. Two people with the same country and language settings are supposed to see the same ranking. This is a first-party policy claim, not an independently audited no-logs report.

A documented HTTP API returns titles, URLs, and query-dependent snippets in JSON or XML from Mojeek's web index. Parameters include country/language boost or restrict, date windows, Safe Search, clustering, and include/exclude lists of up to 25 domains. Plans are pay-as-you-go (contact for access); a limited free trial is offered. Do not treat the API as a licence to the underlying page content: the company says customers must respect publishers' copyright.

Focus lets a user define a set of sites to search, refine includes and excludes from the results page, and save or share that set from a dashboard. No login is required. It is aimed at people who want a portable vertical over the independent index (the company traces the idea to Personal Search in 2006). It is not a full site-search appliance you self-host.

Search Summaries can sit beside web results or on a Summary tab, with citations back to ranked URLs. The April 2024 announcement described a RAG pipeline: query the Mojeek index, then send the query and top results to Mixtral hosted via Lepton. Users can turn the feature off. Treat the LLM hop as a separate data path until Mojeek publishes a current processor list.

At a glance

HQ / register
Mojeek Limited, England (06918197), registered office Croydon
Product start
Public engine from 2004; company incorporated 28 May 2009
Index (vendor)
About timeline: passed 9 billion pages in 2025
Crawler
MojeekBot (own crawler, public IP list)
Source model
Proprietary hosted search (not open source, not self-hosted)
Commercial model
Free consumer search with query/location ads; pay-as-you-go Search API

Best fit when

  • Teams that need organic web results from a crawler-based index that is not Google or Bing
  • Privacy-conscious default search where two users with the same country and language settings should see the same ranking
  • Product or research teams that need a documented JSON/XML search API over an independent index
  • Specialists who want account-free Focus sets over a defined list of sites
  • Site owners already seeing MojeekBot and evaluating robots.txt / indexing behaviour

Poor fit when

  • Organisations that need Google- or Bing-complete coverage and freshness as the only search box
  • Image-first use cases (Images tab uses Pixabay/Openverse, not the independent HTML index)
  • Buyers that require a published DPA, ISO 27001, SOC 2, or independent no-logs audit before shortlist
  • Open-source or self-host mandates (stack is proprietary and hosted by Mojeek)
  • Policies that forbid any US processor, if optional summaries, Stripe billing, or image providers are in scope

Consider instead when

  • When: You need incumbent coverage and ranking quality more than index independence

    Consider: Google Search or Microsoft Bing

    Mojeek's own About timeline puts the independent index at nine billion pages (2025 vendor claim), which is a different class of corpus.

  • When: You need an EU (not UK) legal entity and can accept a different index model

    Consider: Qwant (France)

    Confirm on Qwant's current docs whether web results are still mixed with a US syndicate. Mojeek is UK-incorporated, not an EU company.

  • When: You want a privacy frontend and are willing to send queries to Bing or Google under another brand

    Consider: Startpage or Ecosia

    Those products typically do not replace the incumbent index. Re-read their current provider disclosures.

  • When: You need an independent index but not a European operator

    Consider: Brave Search

    Brave also claims its own index. It is not a UK or EU legal seat.

Jurisdiction & ownership

Legal entity
Mojeek Limited (Companies House 06918197), 74 High Street, Croydon, England CR9 2UU
US parent / control
No known US parent
CLOUD Act exposure (indicative)
Medium
Hosting / residency
Primary web search: Mojeek-managed servers in a dedicated room at a UK green data centre (2018 blog names Custodian, Kent; current About page does not rename the facility). Backup/DR not published. Image search: Pixabay (default) and Openverse. Optional Search Summaries (April 2024): Mixtral via Lepton (US inference host). API billing: Stripe. Newsletter: Buttondown. No public subprocessor list.

UK company, British PSC (Marc Smith, 50-75% voting rights), no known US parent. English law. CLOUD Act score is medium because of named US-group paths (Lepton for optional summaries, Stripe, image providers), not because of ownership. Indicative only, not legal advice.

  • Independent security / no-logs auditNot found
  • ISO 27001Not found
  • SOC 2 / SOC 3Not found
  • GDPR / EU data protectionVendor claimed
  • +3

Considerations & known limitations

  • MediumSmaller, independent index than Google or Bing

    Vendor timeline claims nine billion pages (2025). Expect missing or slower pages versus incumbents. Poor fit as the only search box for users trained on Google completeness.

  • MediumNo public independent no-logs or security audit

    No-tracking and no-IP-in-logs rest on the 2022 privacy policy. Logs are kept indefinitely in de-identified form. There is no published audit PDF to verify that claim.

  • MediumImage search, optional summaries, and billing leave the UK stack

    Pixabay and Openverse handle image queries. April 2024 summaries sent query plus top results to Mixtral via Lepton (US). Stripe bills the API. Do not describe the whole product as UK-only data.

  • MediumNo public DPA, ISO 27001, or subprocessor register

    Enterprise buyers must obtain contract artefacts directly. Backup and DR locations are unpublished.

  • LowColocation facility named only in 2018

    Current About page still claims a UK green data centre and dedicated room. The Custodian/Kent name is from a 2018 blog. Reconfirm the live facility before a residency memo.

Open questions for due diligence

  • Will Mojeek sign a B2B DPA and publish a current subprocessor list (including the live LLM host for Search Summaries)?
  • Is Custodian in Kent still the sole colocation site, and where are backups stored?
  • Is Mixtral via Lepton still the Search Summaries backend in 2026?
  • Has any independent party audited the no-IP-log implementation?
  • Are yearly transparency reports (promised in the content policy) published, and where?
Request vendor DPA / enterprise contact
Ask an EU migration specialist
Visit WebsiteView alternativesFeedback & corrections

Frequently Asked Questions

Yes, according to Mojeek. The Why Mojeek page states that web results are 100% independent and come from MojeekBot plus Mojeek's own ranking. Webmaster documentation describes crawl behaviour, robots.txt handling, reverse-DNS verification under the mojeek.com domain, and a JSON list of crawler IPs. The technology page says the stack was written from scratch in C and is owned by Mojeek Limited. That is a strong first-party claim with public crawler artefacts. It is not a third-party audit that every result is free of other indexes.

No. A July 2023 company blog says the Images tab uses third-party providers. Pixabay was the default. Openverse (openly licensed media) was added as another source. Bing had been a non-default image backend and was removed. Content policy also states that Mojeek indexes HTML webpages and does not index images or video in the web index. If image search is in scope, evaluate those providers separately.

Ordinary web search is answered from Mojeek's index. If you enable Search Summaries, the April 2024 blog says the query and extracted top results are sent to an LLM (then Mixtral via Lepton) to generate a cited summary. Lepton is a US inference host. The feature is optional and can be turned off. Confirm the current model and host with Mojeek before treating summaries as staying on UK infrastructure.

No public B2B data processing agreement or subprocessor register was found on mojeek.com. Documented third-party paths include Pixabay and Openverse for image search, Stripe for API billing, Buttondown for the newsletter, and (as of April 2024) Lepton for optional summaries. Primary web search is described as running on Mojeek-managed servers in a UK data centre. Ask for a current DPA and processor list before a production integration.

No. The technology page states that all technology and IP are fully owned by Mojeek Limited and that the engine was developed from scratch without pre-existing search or crawler technology. There is no official self-host product. Organisation Search lets a customer add, remove, and refresh its own domains in Mojeek's hosted index. It is not a drop-in replacement for a self-hosted crawler.