A search engine with its own index and no ads.
Carpathian Search runs its own crawler and its own index, and ranks pages by how well they answer your search before anything else. Among pages that match about as well, the ones without ad networks, popups, cookie walls and trackers rank higher. The same results are a Search API for your apps and AI agents, and site owners see how their pages show up in Search Engine Console.
curl https://search.carpathian.ai/v1/search \ -H "Authorization: Bearer $CARPATHIAN_SEARCH_KEY" \ -d '{"query": "sourdough starter feeding", "limit": 5}'
{
"query": "sourdough starter feeding",
"intent": "learn",
"results": [
{
"rank": 1,
"title": "Keeping a starter alive",
"url": "https://bread.example/starter",
"snippet": "...feed the starter once a day...",
"published": "2024-03-02"
}
]
}How results are ranked
Relevance carries the largest weight for every search. The other signals reorder pages of similar relevance and never lift an unrelated page above a matching one.
Relevance first
Each result is scored on how well the page body, title and the links pointing at it match your words, and how close together those words sit. Each search is read as a question, a site by name, something recent, a broad topic or something to learn, and the weights change with it, so a navigational search puts the named site's front page first.
Clean pages rise
Among pages that match about equally, a page without ad networks, popups covering the content, cookie walls, trackers and affiliate links ranks above one that has them. Pages light on scripts, with an author, a date and outbound citations, carry more weight than pages built around filler.
Opinion searches favor people
A search with words like best, vs, review or worth it gives more weight to first-hand accounts and discussion threads, and a stronger penalty to affiliate listicles. Every result has a Why here link that shows each signal behind its place, so the ordering can be checked instead of taken on trust.
What it records
Searching needs no account, and the search page loads nothing from another company. It does record what is searched, and the engine's privacy page says so in full.
Carpathian Search records your searches, the results you were shown and the results you open, along with your IP address, browser and device, your language, the site that sent you, and an approximate location at the level of country, region and city. If you are signed in to Carpathian in the same browser, your searches are tied to that account.
It uses those records to rank results better, to watch for abuse, automated traffic and searches for illegal material, and to show site owners how their sites perform as counts, never as who searched. The records stay on servers Carpathian controls, are never sold, and are never shared with advertisers.
The site sets one first-party cookie, cs_vid, holding a random identifier that expires 30 days after your last search. In the European Economic Area, the United Kingdom and Switzerland it is set only if you choose Accept, and a browser sending Global Privacy Control never gets it.
Search Engine Console for site owners
Add your site in Carpathian, prove you own it with a DNS TXT record or a meta tag, and Search Engine Console shows the searches your pages appeared for, with impressions, clicks, click-through rate and average position. A verified site covers every host under it, so www and every subdomain come along.
It also shows which of your pages are indexed and why the rest are not, how each page scores on the signals the ranking weighs, and thirty days of crawling. Sites have no charge.
- Proof of ownership by DNS TXT record or meta tag
- Queries, pages, countries and devices, compared with the period before
- Sitemaps, recrawls and removals queued from the dashboard
- Counts of who found you, never who searched
- PerformanceImpressions, clicks, click-through rate and average position, by query, page, country and device
- Page qualityEach page's scores on the signals the ranking weighs, with advice for the weak ones
- CoverageWhich pages are indexed, and why the rest are not
- Crawl statsThirty days of fetches, status codes, bytes and response time
- IndexNowA key and the requests that tell the engine which pages changed
- ActionsSubmit a sitemap, crawl pages again, or remove a page from results
A Search API for apps and AI agents
A Search API key, starting csa_, calls two endpoints: one returns ranked results with a title, URL, snippet and published date, and one returns the main text of an indexed page without navigation, ads or footers. That is the pair an AI agent needs to search the web and read what it found.
Each key has its own requests a minute and a day, and past either the API answers 429 with a time to retry. The key is shown once, Carpathian keeps only its hash, and a rotated or deleted key stops working within a minute. The API reference, with tool definitions for function calling, is also served as llms.txt for your coding assistant.
What it does not do
The index is Carpathian's own, so its coverage differs from the engines you are used to. Try the searches you run every day before you switch.
It returns web pages, not image, video or shopping results, and it is not a no-logging search engine: searches, addresses and coarse location are recorded as described above.
The Search API reads only pages already in the index. It never fetches a URL because you asked for it, so a page published this morning answers not indexed until the crawler reaches it.
Questions we get asked
Is Carpathian Search free?
Where do the results come from?
Does it show ads?
What does it record about my searches?
Can I make it my browser's default search engine?
How do I get my site crawled?
Can my AI agent use it?
Find it, then get there
Carpathian Maps answers walk, bike, drive and transit directions from the same account that holds your Search API keys, and Carpathian AI hosts the models that can call both.
Search the web without the ads.
Search at search.carpathian.ai with no account. If you run a site, create an account and add it to Search Engine Console to see the searches it shows up for, and the same account makes Search API keys for your apps and agents.