Kolibri gives organizations a German-English LLMA large language model — the neural network behind tools like Claude and ChatGPT, trained on huge amounts of text to predict what comes next.Full definition → they can deploy on their own infrastructure. Aleph Alpha's release announcement says the full model weights are available under the Apache 2.0 license and that its size lets customers run it on their own infrastructure, so internal data does not go to a third-party inferenceRunning a trained model to get answers — the phase where AI is actually used, as opposed to trained.Full definition → service. This is the concrete deployment choice behind its sovereignty claim: customers can control where the model runs.
The company reports 78.1 billion total parameters, with 3.46 billion active per token. Only part of the model participates in each step of generating text, so the active count should not be mistaken for the size of the whole model.
Terms in this piece · Glossary
LLM — A large language model — the neural network behind tools like Claude and ChatGPT, trained on huge amounts of text to predict what comes next.
inference — Running a trained model to get answers — the phase where AI is actually used, as opposed to trained.
open weights — A model whose trained parameters are published for anyone to download and run — unlike API-only models you can access but never possess.
mixture-of-experts — A model built from many specialist sub-networks where only a few activate per token, giving big-model capability at small-model running cost.