Technical Information

The Frequently Asked Questions cover common issues. See our Research collection for an in-depth look at the architecture, training, data mix, and evaluation results. Deployment & compliance information follows:

Apertus Usage Policy

Terms and conditions for model use

Swiss AI Charter

The internal constitution of Apertus

EU Public Summary

Public summary for EU AI Act compliance

EU Code of Practice

Code of practice documentation

Download the official swiss-ai releases on Hugging Face.

Hugging Face

Apertus 8B

8 billion parameters · Apache 2.0 license

Download → Base or Instruct
Hugging Face

Apertus 70B

70 billion parameters · Apache 2.0 license

Download → Base or Instruct

'Instruct' versions are further trained for chat support.

LM Studio

Quick inference on Mac, Linux and Windows.

Over 50 others at AlternativeTo ...

Desktop

vLLM

Recommended for self-hosted inference servers.

Use SGlang for advanced scaling needs.

Server

Community builds exist in formats like MLX and GGUF. Please be aware if you are using a downstream version of Apertus, and seek technical support from the publisher in case of issues.
Contact us in case of security, legal, or other concerns.

Codebase

Swiss AI on GitHub

Training code, evaluation scripts, and model artifacts for Apertus and related projects.

Pre-Training Data

Scripts for reconstructing the training dataset.

Apertus Format

Python library for the custom chat format.

Fully open training infrastructure. We look forward to hear about your efforts to build on our models!