Open-source SerpApi alternatives

SearXNG is a free internet metasearch engine. It sends your query to many search services and databases, merges the results into one page, and does so without tracking or profiling the people who use it. Topics list engines such as Google, Bing, Brave, DuckDuckGo, Qwant and Startpage among the sources it can combine.

The documentation is organized for both users and administrators. Users get search syntax, a list of configured engines and an explanation of why a private instance matters. Administrators get settings in settings.yml, engine configuration, plugins, a bot-protection limiter, optional Redis or Valkey storage, and installation guides covering containers, an installation script and step-by-step setups with Granian or uWSGI behind NGINX or Apache. A long catalog of engines extends beyond web search to sources such as arXiv, Crossref, GitHub and Mastodon.

SearXNG is written in Python under the AGPL-3.0 license. Because it is meant to be self-hosted, you can run your own instance for yourself or a group; many volunteers also publicly host instances. It appeals to people who want to reduce reliance on any one search engine and keep their queries private.

Wigolo is a local-first web access layer for AI agents. It exposes search, page fetching, crawling and research capabilities over the Model Context Protocol, so a coding agent can look things up on the web without you buying a search API key or sending queries through a metered cloud service.

The README says it works with a long list of clients, including Claude Code, Cursor, Codex and Gemini CLI as well as OpenCode, Windsurf, Zed and VS Code, and with frameworks such as LangChain, CrewAI, LlamaIndex, the Vercel AI SDK and n8n, plus any MCP client or plain REST. Repository topics describe it as a metasearch and web crawler and name Exa, Firecrawl and Tavily as the services it stands in for. The project is in public beta.

Wigolo is written in TypeScript and its license is listed as 'Other' on GitHub, so read the licensing section before commercial use. It runs on your own machine. It suits developers who want agent-ready web search that is private and free of per-query costs.

OpenSERP is a free, open-source search results API and command-line tool. With one command it provides live, structured results from Google, Yandex, Baidu, Bing, DuckDuckGo and Ecosia on localhost, without API keys or per-search billing, which makes it useful as a search tool for LLMs and agents or as a backend for SEO rank tracking.

Each of the six engines has its own endpoint using the same JSON schema, and a megasearch endpoint runs one query across several engines, merging and deduplicating results. A URL extraction call returns search results together with clean Markdown of the target pages. It captures SERP features such as AI summaries, answer boxes and related searches, supports image search and filters, and offers JSON, Markdown, text and NDJSON output, plus proxies, caching and a resilient mode.

OpenSERP is written in Go and released under the MIT licence, with prebuilt Docker images and go install options. Official SDKs for JavaScript and others work against either a self-hosted server or the hosted API, which the README offers for people who would rather not run infrastructure. It suits developers building agents, research tools and rank trackers.

GoogleScraper is a Python module built to scrape results from several search engines, including Google, Bing, Yandex and DuckDuckGo, with support for asynchronous networking. It was created for developers who needed programmatic access to search engine results pages for research, SEO or data collection purposes.

According to its own README, the project is no longer actively maintained by its author as of 2019, who explains that Python and Selenium-based scraping were a poor match for how modern bot-detection and browser automation work, in contrast to Node.js tools built on Puppeteer. The author's related, actively developed successor project moved to a general-purpose crawling infrastructure built in JavaScript, and a commercial scraping service called Scrapeulous is also mentioned as an alternative for people who want a managed option.

GoogleScraper itself remains available as open-source Python code, written with an HTML/Python stack, and is released under the Apache-2.0 license. New projects should weigh the author's stated preference for the newer JavaScript-based tooling before building on this module.