SearXNG is a free, self-hostable metasearch engine that aggregates results from more than 270 search services and databases, and it is meant for people and organizations that want web search without being tracked or profiled.

What it is

SearXNG is a metasearch engine written in Python. Instead of crawling and indexing the web itself, it queries other search services and databases, merges what they return, and presents a single result set through its own interface. The project describes itself plainly: users are neither tracked nor profiled. It is licensed under the GNU Affero General Public License (AGPL-3.0), and its documentation lives at docs.searxng.org.

The concrete problem it solves is dependency on a single commercial search provider that logs queries and builds profiles. SearXNG sits in the search-engine ecosystem alongside the general-purpose web search engines named in its topic list — Google, Bing, DuckDuckGo, Brave and Qwant — and it replaces the direct use of any one of them. Because the engine runs on hardware the operator controls, the query stream stays with the operator rather than with an advertising-funded intermediary. The topic list also carries the "degoogle" label, which points at the same motivation: reducing reliance on a single dominant provider.

Key capabilities

• Metasearch aggregation across more than 270 sources, combining results from multiple search services and databases into one result page.
• No user tracking and no profiling, as stated directly in the README.
• A self-hosted deployment model written in Python, so the operator controls the instance and the query path.
• Configuration through a dedicated settings system, documented in the Configuration guide at docs.searxng.org.
• A web front end, with the brand asset shipped in the source tree at client/simple/src/brand/searxng.svg.
• Community localization through Weblate on translate.codeberg.org, under the SearXNG project.
• Community support through the Matrix room #searxng:matrix.org.

Who uses it and how

• Privacy-conscious individuals who run their own instance rather than sending queries to a commercial engine.
• Organizations with data-handling obligations that cannot accept third-party query logging as part of their search workflow.
• People following the "degoogle" pattern described in the topic list, who use SearXNG as the everyday replacement for a direct Google or Bing session.
• Self-hosters and homelab operators; the project's 33,794 forks relative to 37,278 stars indicate a codebase many people copy to deploy their own instance.
• Contributors and translators, who work through the Matrix room and the Weblate translation project.

Getting started

The README does not carry install commands; it points to the Installation guide at docs.searxng.org/admin/installation.html for setup and to the Configuration guide at docs.searxng.org/admin/settings/index.html for fine-tuning. Both guides are the authoritative source for the exact deployment method.

How it compares

The facts provided list no paid products that SearXNG replaces, so no licence, hosting or cost comparison can be drawn here. Among the tools the facts do name, SearXNG sits in the same lineage as the SearX project, which appears in the topic list. Its distinguishing positions against the engines it queries — Google, Bing, DuckDuckGo, Brave and Qwant — are self-hosting and the absence of tracking and profiling, not a different index of the web.

When to use it — and when not to

Choose SearXNG when running and configuring a Python service is acceptable and keeping queries on your own infrastructure matters. The README is short and defers almost everything to external documentation, so anyone needing a step-by-step deployment walkthrough in the repository itself should expect to read the docs site instead, and the 215 open issues show active unresolved work. A self-hoster must be prepared to operate the service and tune it through the settings system, since the facts describe no hosted option and no managed alternative.