Insights & Updates

A deep-dive into how floating point numbers work, what quantization actually does under the hood, and why it lets you run large language models on your laptop.

We're launching openoranje to bring open-source Dutch language models to the Netherlands, focusing on privacy-first edge inference.

A deep dive into the unique challenges we face when training language models for Dutch, and how we're solving them.

A practical guide to deploying openoranje models on edge devices for privacy-preserving AI applications.