Skip to main content

Documentation

What are the differences between the search engine providers?

Compare web and image search providers, how Uruky merges results, keyword vs semantic resolution, how query type should guide provider choice, and how to pick an order that fits you.

Uruky does not rely on a single search index. You choose an ordered list of web and image search providers in your settings. When you search, Uruky queries providers in that order; if one returns too few results, it tries the next, and so on. When it helps latency, the first two providers (and Uruky Site Search, when enabled) may be queried in parallel. Results are then merged in a round-robin fashion and adjusted for your domain boost/exclude preferences.

Uruky does not offer Generative AI/LLM-based features. We also cannot control how partners build their products or what tooling they use internally; our commitment is about Uruky's product surface and which API modes we call.

How providers are used

For each search, Uruky starts with your top-ranked enabled providers. If they do not return enough results (only Mojeek, Mwmbl, and Uruky Site Search report an exact total; for others we estimate), later providers are queried in order. When it helps latency, the first two providers (and Uruky Site Search, when enabled) may be queried in parallel. Final results interleave hits from each provider (first from provider 1, first from provider 2, second from provider 1, and so on), then domain pins, boosts, and exclusions are applied.

Keyword matching vs semantic search

How a provider resolves your query differs. Some use keyword matching (looking for the terms you typed); others appear to use semantic search (matching meaning and intent more broadly). That changes how literal your query needs to be.

  • Uruky Site Search, EUSP, Mojeek, Marginalia, Mwmbl, and Linkup (in the "fast" mode Uruky uses) use keyword matching.
  • Serper and PriEco appear to use semantic search.

Web search providers

Mojeek

UK-based independent search engine with its own crawler and index. Default first provider. Tends toward European and smaller-web results rather than mainstream Google-style rankings.

Keyword/index results without LLM interference in the search path Uruky uses.

EUSP (Ecosia / Qwant)

EU-based results via the European Search Perspective API. Native markets are French, German, and English; other languages fall back to English.

Keyword/index results from the Staan web-search index, separate from Ecosia/Qwant consumer AI features.

Linkup

Real-time web search API (EU). A strong choice if you want fresher, broader web coverage closer to what you expect from mainstream engines.

Uruky uses Linkup's "fast" depth: index retrieval without LLM query interpretation, not their agentic or answer modes.

Marginalia

Indie search (EU-based) focused on the small and independent web. Excellent for niche, text-heavy, or non-commercial queries where mainstream indexes underperform.

Keyword/index results without LLM interference in the search path Uruky uses.

Serper

UK-based API that returns Google-style organic results without you visiting Google directly. More private than using Google yourself, but not an independent index. Available by choice for users who want familiar result shapes.

PriEco

Private, eco-friendly, independent, and open source web search engine, based in the EU.

It's important to note that PriEco uses Cloudflare for DDoS, so technically, there is a layer of non-EU-based privacy protection for the web search requests.

Mwmbl

Open-source, non-profit, community-supported search engine with its own index (EU/UK-based). A good fit for independent and smaller-web queries alongside Marginalia and PriEco.

Uruky Site Search

Our focus for Uruky Site Search is the small and indie web: websites that do not play by mainstream SEO rules and often struggle with discoverability on big indexes.

Optional add-on: when enabled, it queries our (tiny) index built from opted-in Site Search domains and our curated indie and small-web URL list, in parallel with your web providers. See Uruky Site Search .

Keyword/index results without LLM interference in the search path Uruky uses.

Image search providers

Image search uses a separate provider list:

Pixabay

EU-based royalty-free stock images, which works really well for specific niche queries.

Pixabay allows AI-generated images when contributors tag them, but their public API does not expose or filter that flag, so Uruky cannot exclude them at query time.

Serper

Same UK-based provider as for web search. Works better for more generic/standard queries

Matching providers to your query

No single search index excels at every kind of query. Your default provider order is a good baseline, but the best results often depend on how specific your query is, what topic you are searching for, how fresh the information needs to be, which language or region you care about, and whether you are searching the web or images (each uses a separate provider list).

  • Broad or mainstream queries — head terms, current events, products, well-known people — usually work best with Serper or Linkup, which offer fresher, broader coverage closer to mainstream engines.
  • European-centric general web searches often work well with Mojeek, one of the world's largest independent indexes — not a small-web specialist, but a strong default for EU-oriented queries.
  • Niche, indie, text-heavy, or non-commercial queries often surface better on Marginalia, PriEco, or Mwmbl, which focus on the small and independent web.
  • EU-centric searches in English, French, or German can benefit from EUSP, which draws on the European Search Perspective API (those are its native markets; other languages fall back to English).
  • Queries about the small web, indie sites, or opted-in Site Search domains can benefit from Uruky Site Search when enabled: it targets pages mainstream engines under-rank or miss (sites that do not optimize for SEO) by crawling opted-in customer domains and a curated indie/small-web URL list in parallel with your web providers (the index is still tiny but growing).
  • For image search, Pixabay tends to work well for specific or niche stock-style queries, while Serper is stronger for generic or mainstream image lookups.

When results feel off for a particular search, use the Try with: links on the results page to re-run with a different primary provider — no settings change needed.

Choosing the right order

By default, Uruky prioritizes independent and European-centric indexes. If you are used to Google, Bing, or DuckDuckGo, try putting Serper or Linkup first in your web search provider list.

You can download the recommended settings preset and import it, or apply it automatically.

Per-search override: use the sp query parameter with a comma-separated provider list (e.g. sp=serper,linkup ).

Related