Private Inference

1 min read Original article ↗

Tinfoil inference API

Our SDKs provide secure, verified access to our inference API. Each API connection is automatically verified and encrypted directly to an attested secure enclave, and is compatible with the OpenAI API standard, making it a drop-in replacement for most existing deployments and workflows.

Everything you need to build

Production-ready inference with the features your app needs, and privacy built into every request.

OpenAI-compatible

Point your existing OpenAI client at Tinfoil and keep your code. No rewrites required.

Verified & encrypted

Every request is encrypted directly to an attested secure enclave, verified client-side.

Tool calling

Let models call your functions and external APIs to build agents and workflows.

Vision & images

Analyze and reason over images and screenshots with multimodal vision models.

Document processing

Extract and analyze content from PDFs and long-form documents.

Structured outputs

Get JSON responses that conform to your schema on every call.

Inference code examples

Pick a model to see a ready-to-run code sample. Each request is automatically verified and encrypted to a secure enclave, so you get private inference without changing how you build.

Supported AI Models