Solutions
Face Recognition Model Licensing
Commercial licensing for InsightFace's open-source model packages (buffalo_l, antelopev2, buffalo_s, buffalo_m, etc.) and proprietary high-accuracy closed-source models. From academic research to full commercial deployment.
Key Features
Face Recognition Model Licensing
Open-Source Model Commercial License
Obtain commercial usage rights for model packages from the InsightFace open-source project, including buffalo_l, antelopev2, buffalo_s, buffalo_m, and more. Code is MIT licensed; models require separate commercial licensing.
Closed-Source High-Accuracy Models
Access proprietary models with higher accuracy than open-source variants, optimized for demanding production environments requiring the best possible face recognition performance.
NIST FRVT Top Performer
Our models consistently rank among the top performers in NIST Face Recognition Vendor Test (FRVT), the gold standard for face recognition evaluation worldwide.
Flexible Licensing Terms
Licensing options tailored to your deployment scale and use case — from single-product licenses to enterprise-wide deployment rights with dedicated support.
Code Example
Compare two face embeddings with buffalo_l
Use the buffalo_l package to extract normalized feature vectors from two face images and compute cosine similarity for verification and matching workflows.
1import cv22import numpy as np3from insightface.app import FaceAnalysis45app = FaceAnalysis(name="buffalo_l")6app.prepare(ctx_id=0)78img1 = cv2.imread("person_a.jpg")9img2 = cv2.imread("person_b.jpg")10if img1 is None or img2 is None:11 raise FileNotFoundError("input image not found")1213faces1 = app.get(img1)14faces2 = app.get(img2)15if not faces1 or not faces2:16 raise RuntimeError("face detection failed")1718feat1 = faces1[0].normed_embedding19feat2 = faces2[0].normed_embedding20similarity = float(np.dot(feat1, feat2))2122print("face 1 feature:", feat1[:5])23print("face 2 feature:", feat2[:5])24print("cosine similarity:", similarity)Evaluate our highest-accuracy proprietary models on your data
Beyond the open-source packages, our closed-source face recognition models deliver materially higher accuracy on demanding production scenarios. To make procurement decisions easier, qualified enterprise customers can run a structured evaluation against the same proprietary models we deploy in production.
Structured evaluation access
Qualified enterprise teams receive private API access (or on-premise evaluation instructions) for a scoped evaluation period agreed during scoping.
Production-grade proprietary models
Test against the same closed-source models that consistently rank among the top performers in NIST FRVT — not a downgraded preview build.
Bring your own benchmark
Run identity verification, KYC, 1:N search, or access-control workloads on your internal datasets and compare directly with your current vendor or open-source baseline.
Engineering support included
Our team helps with SDK integration, threshold tuning, and result interpretation throughout the evaluation so you can finalize a commercial licensing decision with confidence.
Evaluations are scoped per engagement and intended for enterprise procurement. Submit an enterprise inquiry with your use case, expected request volume, and target deployment to start the process. Eligibility, evaluation duration, and access modality are subject to review.
Start Private Model EvaluationUse Cases
Use Cases
Ready to deploy the world's best face AI?
Whether you need face recognition model licensing, face swapping commercial rights, InspireFace SDK, or custom AI cooperation — reach out to get started.