Go EuropeAIdirectory

Evidence-backed European AI register

Danish Foundation Models

Reviewed identity, roles, model listings, pricing and evidence for Danish Foundation Models.

Dataset compiled 2026-08-28 · evidence over inference

HQ Europe: Unreviewedconsortiummodel labnational initiative6 model listings0 token-price rows

Reviewed identity

Headquarters scope
headquarters unknown
Headquarters country
unknown
Headquarters city
unknown
Legal entity
open collaboration across Danish universities and research institutions — no single legal entity
Ownership
public research consortium
Founded
2021
Confidence
medium
Last verified
2026-08-22

Links

Official website

Headquarters evidence

Open interactive profile

Headquarters evidence

No qualifying headquarters passage was promoted.

Consortium with no single legal entity and no qualifying headquarters wording verified.

Evidence boundaries

Organisation listings identify a reviewed canonical record. Headquarters is independent of incorporation, registered offices, ownership, hosting and runtime. Model listings show an association in the source register; they do not by themselves establish model creation, current API availability, hosting location or infrastructure ownership.

Model listings

  1. Munin 1.0 — munin-qwen3.5-9Bcatalogued with this organisation · origin not asserted
    Type
    text
    Parameters
    10B
    Context
    unknown
    Languages
    Danish, English
    Licence
    Apache-2.0 (per DFM release note; base-model terms may also apply — check the card)
    Release
    2026-06-11
  2. Munin 1.0 — munin-apertus-8bcatalogued with this organisation · origin not asserted
    Type
    text
    Parameters
    8B
    Context
    unknown
    Languages
    Danish, English
    Licence
    Apache-2.0 (post-trained on Apertus, the Swiss open model)
    Release
    2026-06-11
  3. Munin 1.0 — munin-ministral3-8Bcatalogued with this organisation · origin not asserted
    Type
    text
    Parameters
    9B
    Context
    unknown
    Languages
    Danish, English
    Licence
    Apache-2.0 per release note (Ministral base terms may apply)
    Release
    2026-06-11
  4. DFM-Mimircatalogued with this organisation · origin not asserted
    Type
    text
    Parameters
    1–2B (reported as 1B; HF org lists 2B)
    Context
    unknown
    Languages
    Danish
    Licence
    unknown
    Release
    2026
  5. munin-7b-alphacatalogued with this organisation · origin not asserted
    Type
    text
    Parameters
    7B (continual pretraining of Mistral-7B-v0.1 on Danish Gigaword)
    Context
    8,192 tokens
    Languages
    Danish
    Licence
    Apache-2.0
    Release
    2024-01-11
  6. CROCO variants (based on Munin Apertus 8B)catalogued with this organisation · origin not asserted
    Type
    text
    Parameters
    8B
    Context
    unknown
    Languages
    Danish
    Licence
    unknown
    Release
    2026

Profile note

Denmark's national open-model programme, run by five institutions: University of Southern Denmark, Aarhus University, University of Copenhagen, the Alexandra Institute and the Center for Humanities Computing Aarhus. Covers the whole pipeline — training data, models, evaluation benchmarks and 20+ industry use cases. The Munin 1.0 family (11 Jun 2026) marks a strategic shift: instead of pretraining from scratch, DFM now post-trains several best-in-class open bases (Qwen 3.5, Apertus, Ministral) for Danish — pragmatic, and it means base-model licence terms need checking per card. DFM-Mimir is th…

Record sources

  1. https://www.foundationmodels.dk/
  2. https://www.foundationmodels.dk/news/
  3. https://huggingface.co/danish-foundation-models
  4. https://huggingface.co/danish-foundation-models/munin-7b-alpha
  5. https://www.foundationmodels.dk/blog/2024/01/11/releasing-munin-7b-alpha---a-danish-llm.html
  6. https://chc.au.dk/research/danish-foundation-models-dfm
  7. https://techsavvy.media/en/new-danish-ai-outperforms-the-major-models-in-danish-and-english/