Kolibri Has Landed: A Sovereign Open-Weight Model
Aleph Alpha released Kolibri, an Apache 2.0 English-German MoE with 78B total and 3B active parameters.
Aleph Alpha released Kolibri, an English-German mixture-of-experts transformer with 78 billion total parameters, 3 billion active, and a context window up to 1 million tokens. Full weights are on Hugging Face under Apache 2.0, following Kolibri Origin, a 30B-total, 3B-active model with a 65k context. Reported scores include AIME 2025 at 96.9, GPQA Diamond at 84.3, and LiveCodeBench v6 at 85.9, competitive with models that use up to four times as many active parameters. It is aimed at on-premise sovereign deployments in German public administration, industrials, and aerospace.
- Kolibri is a 78B-total, 3B-active MoE with up to 1 million tokens of context.
- Weights are on Hugging Face under the Apache 2.0 license.
- AIME 2025 score is 96.9; GPQA Diamond is 84.3.
- Built for on-premise sovereign use in German government and industry.
- It succeeds Kolibri Origin, a 30B model with a 65k context window.
- Vendors
- Aleph Alpha
- Organizations
- Aleph AlphaHugging Face
- Countries
- Germany
From the original article
14 min read at aleph-alpha.com
On the Day of German Reunification, we are releasing our new model: Kolibri. Kolibri is an English-German Mixture-of-Experts Transformer with 78B total parameters, 3B active. It supports context lengths of up to 1M tokens.
By bastitx · aleph-alpha.com
A short excerpt; the article and its images belong to the publisher. Summary, key facts and assessment above are AI-written from the article and can be wrong.