Self-hosted SERP API that unifies Google, Bing, Baidu, and the rest

What it does OpenSERP is an open-source API and CLI written in Go that queries Google, Yandex, Baidu, Bing, DuckDuckGo, and Ecosia, then returns normalized results in JSON, markdown, text, or NdJSON. It is built to be self-hosted, with configurable proxy pools, caching, and a resilient mode for when search engines throw captchas. Beyond raw links, it extracts AI summaries, answer boxes, related searches, and can even fetch target-page content as clean markdown for LLM grounding.

The interesting bit The megasearch endpoint is where it gets clever: it queries multiple engines in parallel, deduplicates results by normalized URL, and clusters the occurrences so you can see how each engine ranked the same page. There is also a fast mode that simply hits the quickest engine and a sequential fallback mode when you care more about reliability than breadth.

Key highlights

• Dedicated endpoints for six search engines plus image search across all of them.
• Cross-engine aggregation with deduplication and per-engine rank clustering.
• Advanced filtering by language, date range, file type, and site-specific queries.
• Optional URL extraction that turns target pages into markdown or text, embeddable directly under search results.
• Tagged proxy pools and captcha detection with a 503 response when engines rate-limit you.

Caveats

• The README documents a 503 captcha response and proxy support, but is vague on how often rate limits trigger or how many proxy hops are needed for stable production use.

Verdict Worth a look if you are building SEO tools, AI agents, or research pipelines that need structured search data without vendor lock-in. Skip it if you expect fully hands-off operation and do not want to manage proxy infrastructure.