Large Language Models

Access large language models running side Trusted Execution Environments

What it is

Brief description

Learn about TEE security

Available Models

We provide access to leading open source and proprietary models. Where possible, we provide the open source models in confidential mode, so all user prompts are completely private.

As of 30 October 2025, we provide:

1. Qwen (confidential)

2. OpenAI open source models (confidential)

3. Deepseek (confidential)

4. Google Gemma (confidential)

5. OpenAI proprietary models (non-confidential)

6. Anthropic proprietary models (non-confidential)

You can access the full list of supported models via the endpoint.

Using Confidential AI Models

TODO: Provide Lang chain code example