Simple, Transparent Pricing

ModelPassport supports any scikit-learn compatible classification model — logistic regression, random forests, gradient boosting, SVMs, and more. You can upload a pre-trained model as a joblib file, or simply upload your dataset and we'll train a RandomForest baseline for you. The audit focuses on the model's outputs, not its architecture, so most binary and multiclass classifiers are supported.

Your privacy is our top priority. Datasets and model files are processed in memory and deleted immediately after the audit completes — we never store your training data or model weights. Audit results (scores, certificate metadata) are retained and associated with your certificate ID. All API communication is encrypted in transit. See our Privacy Policy for full details.

Certificates do not expire, but they are point-in-time assessments. A certificate reflects the fairness of your model on the data provided at audit time. We strongly recommend re-auditing whenever you retrain your model, update your training data, or deploy in a new context. The EU AI Act and most compliance frameworks expect ongoing monitoring, not one-time certification.

ModelPassport's audit methodology is aligned with: EU AI Act (high-risk system fairness requirements, Article 10 data governance), US EEOC 80% / Four-Fifths Rule (disparate impact in employment decisions), India DPDP Act 2023 (data processing fairness), and the IEEE Ethically Aligned Design framework. Our Gemini Governance layer explicitly maps findings to regulatory requirements and produces compliance-ready language.