Data protection
All data processed by the TabPFN API — including uploads via the/tabpfn/* endpoints and predictions — is handled using secure, compliant infrastructure.
Encryption
- All communication is encrypted in transit using TLS 1.2+.
- Data stored on our servers is fully encrypted at rest.
- Encryption keys are securely managed and rotated in line with industry best practices.
- Services run on hardened cloud infrastructure with strict access control, audit logging, and network segmentation.
- Authentication is enforced via Bearer tokens on every request. All internal systems follow the principle of least privilege.
- Continuous monitoring ensures compliance with modern enterprise security standards.
- No data is shared between customers. Each uploaded dataset, fitted model, and prediction request is scoped to your account and processed in isolated environments.
- Data from one user or organization is never accessible to another.
Model privacy
TabPFN foundation models are never trained or fine-tuned on customer data. All training is performed exclusively on synthetic data under controlled conditions — guaranteeing complete separation between customer data and model weights.Data retention
- Uploaded data may be stored temporarily for reproducibility and caching.
- You can opt out of storage or request data deletion at any time by contacting us.
- We retain only minimal metadata necessary for authentication, billing, and audit purposes.
Enterprise deployments
TabPFN is available through enterprise deployments including on-prem and VPC deployment on AWS SageMaker and Azure AI Foundry, giving you full control over where your data is processed.Azure AI Foundry
Deploy TabPFN in your Azure environment.
AWS SageMaker
Deploy TabPFN in your AWS environment.