LibertAI Search

Private web search
for your agents.

Query Google, Bing and DuckDuckGo in parallel and get one deduplicated, consensus-ranked answer. From $2 per 1,000 queries, with no profiling and no search history tied to you.

Web · News · Images · Academic · one response shape
best private inference API 2026 Consensus · 3 engines
libertai.io/api
LibertAI · The private inference API
GBD0.98
docs.libertai.io/apis
OpenAI-compatible API reference
GBD0.91
news.example.com/private-ai
Confidential inference goes mainstream
GBD0.64
blog.example.com/seo-post
Top 10 AI APIs (sponsored)
GBD0.22

Every query fans out to

GoogleBingDuckDuckGo
One query, three engines, one answer

Search that respects the person asking.

The same privacy guarantees as LibertAI inference, applied to web search. Geo-distributed workers do the searching, consensus does the ranking, and you keep your queries to yourself.

Consensus ranking

Three engines agree, so you don't get gamed.

Every query fans out to Google, Bing and DuckDuckGo in parallel. Results are deduplicated by URL and ranked by cross-engine agreement, so a link all three surface outranks one engine's SEO-gamed top hit.

How ranking works
consensus
All three engines agree
GBD
Two of three
GBD
Single-engine SEO hit
GBD
Private by architecture

Nothing ties the query to you.

Your query reaches the engines from rotating workers spread across three continents, with nothing linking it to your identity. No profiling, no ads, no search history attached to a person.

See how privacy works
workers · rotating
No profiling, no ads
Rotated every 12 hours
No history tied to a person
Built for agents

Clean JSON, ready to ground an LLM.

One POST returns structured, deduplicated results. A companion fetch endpoint returns the readable text of any page, so your agent can cite sources without touching HTML.

API reference
response.json
{
  "query": "private inference API",
  "results": [{
    "title": "LibertAI",
    "url": "libertai.io/api",
    "engines": ["google","bing","ddg"],
    "consensus": 0.98
  }]
}
More in the box

One endpoint, four search types.

Web, news, images, academic

Four search types behind one endpoint. Switch with a single request parameter and keep the same unified response shape.

Pay per query with x402

From $0.002 per query on the same credits as inference. With x402 micropayments an agent can pay for a single search: no account, no card, no key.

Infrastructure that stays fresh

Search workers run as short-lived VMs on Aleph Cloud, health-checked every 30 seconds and rotated every 12 hours for IP diversity and reliability.

Simple per-query pricing

Priced per query, per engine.

Billed on the same credits as the inference API. You choose which engines to query, and only engines that return results are charged.

GGoogle
$0.002
per query · $2.00 / 1,000
BBing
$0.002
per query · $2.00 / 1,000
DDuckDuckGo
$0.002
per query · $2.00 / 1,000

A full consensus query combines the per-engine prices · page fetches & x402 documented in the dev docs · figures illustrative

Quickstart

One POST, structured results.

search.sh
# fan out to all three engines, get one ranked list
curl https://api.libertai.io/v1/search \
  -H "Authorization: Bearer $LIBERTAI_API_KEY" \
  -d '{
    "query": "private inference API",
    "engines": ["google", "bing", "ddg"],
    "type": "web"
  }'

Search, without the surveillance.

Grounding for your agents, priced per query, private by design.