Tell real residential IPs from fakes, each with a risk score and the evidence.
IPBGP scores any IPv4 or IPv6 address for residential trust, connection type, VPN and proxy use, residential-proxy exits, and abuse. Every verdict comes with the signed, human-readable reasons that produced it. We collect the data ourselves and run our own global measurement fleet.
curl https://ipbgp.org/v1/ip/8.8.8.8 \
-H "Authorization: Bearer ipbgp_live_xxx"One JSON response: classification, trust score, evidence, routing, geolocation, and abuse signals.
{
"ip": "8.8.8.8",
"classification": { "connection_type": "hosting", "is_datacenter": true },
"trust": {
"score": 12,
"residential_probability": 0.01,
"recommendation": "block-as-residential",
"reasons": [
{ "sign": "-", "text": "Announced by a hosting / cloud network" },
{ "sign": "-", "text": "No broadband eyeball footprint" },
{ "sign": "+", "text": "Stable prefix, RPKI-valid" }
]
},
"network": { "asn": 15169, "as_name": "Google LLC", "as_type": "content", "prefix": "8.8.8.0/24", "rpki": "valid" },
"geo": { "country": "US", "city": "Mountain View", "confidence_km": 25 },
"anonymization": { "is_vpn": false, "is_residential_proxy": false, "is_tor": false },
"risks": []
}Explainable by design
Every score comes with trust.reasons: signed, neutral evidence you can read, with each item marked positive or negative. Tune your rules on what actually happened, not a bare number.
Residential-proxy detection
Addresses that look residential but are sold as proxy exit nodes are the case most tools miss entirely. We find them by pairing network topology with our own active measurement data.
BGP and AS evidence
Customer-cone size, upstreams, prefix stability, RPKI, anycast. The verdict reasons at the routing layer — a place a geo database simply cannot reach.
First-party intelligence, end to end
We don't resell someone else's database. Routing and registry data flows in continuously; our own global detection nodes handle active measurement; the two feed a single trust verdict with a full evidence trail attached.
- Continuous BGP, RPKI & registry ingestion
- Our measurement fleet probes connectivity & TLS posture
- One verdict: classification, trust score, evidence, risks
- Field depth scales with your plan; the scoring logic never changes
Common questions
What does the trust score mean?+
A number from 0 to 100 estimating how likely an IP is a genuine end-user or residential connection, as opposed to hosting, datacenter, or an anonymizing service. Every score comes with trust.reasons — the signed, human-readable evidence behind it — so you can build rules on what you can actually see rather than a bare number.
How is this different from a plain IP-to-geo database?+
A static database tells you where an address is registered. IPBGP tries to answer what an address actually is — residential, hosting, mobile, VPN/proxy, or a residential-proxy exit — by looking at live routing data alongside our own active measurement. You get an evidence trail for each verdict, not just a city and country.
What is residential-proxy detection?+
Some IPs look like ordinary home connections but are sold as proxy exit nodes — the case most tools get wrong. We find them by pairing network topology data with measurements from our own detection nodes.
Do you support IPv6?+
Yes — full IPv4 and IPv6 coverage, run through the same verdict engine.
How do API keys, scopes and environments work?+
Every request needs a Bearer key. Keys carry scopes (ip:read, bulk:read, dataset:download) and come in live and test variants. The full secret is shown exactly once when you create a key; after that we store only a hash and a display prefix.
What are the rate limits and quotas?+
Both depend on your plan. Every response includes RateLimit-Limit, RateLimit-Remaining, and RateLimit-Reset headers. A 429 adds a Retry-After header so your client knows exactly when to try again.
Can I try it before signing up?+
Yes. There's a live, unauthenticated lookup right on this page. After that, the free tier gives you 1,000 lookups a month with no card required.
How does payment work?+
We accept credit cards, Alipay, and WeChat Pay. For Hong Kong local bank transfer or international wire, contact us.
Do you offer an offline database or self-hosted deployment?+
Neither is a self-serve product. Self-serve plans are API access, and Business adds dataset snapshot downloads. If you need something beyond that, contact us and we'll talk it through.
Start for free
1,000 lookups a month, no card required. Paid plans unlock the full evidence fields, higher limits, and dataset downloads.