Apertus is the fully open Swiss language model, developed by the ETH AI Center and the EPFL AI Center as part of the Swiss AI Initiative. Weights, training data, and recipes are openly available: no black box, no lock-in.
onprem.ai is listed on apertus-ai.org as an inference provider: we bring Apertus onto your own hardware, ready for production and with full data sovereignty.
What is Apertus?
- Fully open: Model weights are distributed through the swiss-ai organization on Hugging Face, including a 70B variant.
- Apertus 1.5: The current release adds image understanding, a thinking mode, and a four times longer context window.
- Swiss values: Trained multilingually and built for transparency and traceability, ideal for regulated environments.
Why does this fit onprem.ai?
Apertus proves that sovereign AI is possible without depending on US hyperscalers. That is exactly what we build infrastructure for: with onprem.ai, Apertus runs in your own data center, from GPU selection to production operation.