Identity checks you can defend in an audit.
IDNexus is a server-authoritative identity assurance API. Every call returns a confidence-scored decision backed by presentation-attack detection, document intelligence and biometric face matching, and every decision keeps the evidence that produced it.
What we verify
The core service proves that the person presenting a document is physically present, is who they claim to be, and is not a spoof, a deepfake or a replay. Around that core we deliver document intelligence and identity attribute validation through the same API.
- Biometric liveness. Active and passive detection that separates a live human from a photo, a screen replay or a 3D mask, scored per frame.
- Face matching. The live face is compared with the face on the submitted document using an ensemble of embedding models. You get a similarity score, not only pass or fail.
- Document intelligence. OCR extracts structured fields from passports, driver's licences and ID cards. Watermark and layout checks detect tampering and low-quality scans.
- Deepfake and injection defence. Enterprise requests are screened for synthetic media, virtual-camera injection and real-time face-swap filters.
How we make decisions trustworthy
Input quality gate
Blur, glare, occlusion and low-light frames are rejected before any model runs. They are never counted as biometric failures.
Server-authoritative decisions
Every decision is computed on the server. There is no client-signed liveness proof, no browser-side trust and no check that a JavaScript edit can bypass.
Versioned result schema
Each result carries the model version, the tier used and the exact scores behind the decision, for audit, dispute resolution and regulator review.
Graceful degradation
If an engine is unavailable, the response reports which engine answered and at what confidence, so you can accept, retry or escalate with full information.
Infrastructure
The engine runs on Google Cloud Run in a hardened Python container. Every call is metered through Cloudflare Workers, so usage, quota and billing are enforced at the edge before a request reaches the model layer. API keys are stored as SHA-256 hashes, passwords use PBKDF2-SHA256 and identity numbers are hashed before storage. The privacy policy sets out the full data-handling specification.
Pricing
Credits are prepaid and priced per call. One credit is $0.01, and unused credits stay in your account indefinitely. Basic, Standard and Enterprise calls cost 10, 50 and 100 credits. See pricing.