Go EuropeAIdirectory

Evidence-backed European AI register

Alexandra Institute

Reviewed identity, roles, model listings, pricing and evidence for Alexandra Institute.

Dataset compiled 2026-08-27 · evidence over inference

research institutemodel lab3 model listings0 token-price rows

Reviewed identity

Headquarters scope
EU headquarters
Headquarters record
Aarhus
Legal entity
Alexandra Instituttet A/S — a research-based Danish non-profit company
Ownership
non-profit; owned by the Aarhus University Research Foundation
Founded
1999
Confidence
medium
Last verified
2026-08-22

Links

Official website

Open interactive profile

Evidence boundaries

Organisation listings identify a reviewed canonical record. 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. mmBERT-small hallucination-detection family (per-language: da, no, sv, nl, it, hu, lt, lv, pl, ro, sk, sl, …)catalogued with this organisation · origin not asserted
    Type
    text
    Parameters
    0.1B each
    Context
    unknown
    Languages
    many European languages, one model per language
    Licence
    unknown
    Release
    2026 (actively updated)
  2. DaNLP model collection (Danish NER, sentiment, embeddings, ASR)catalogued with this organisation · origin not asserted
    Type
    text
    Parameters
    various
    Context
    unknown
    Languages
    Danish
    Licence
    unknown
    Release
    2019–
  3. DaNE dataset (Danish Dependency Treebank annotated with PER/ORG/LOC)catalogued with this organisation · origin not asserted
    Type
    text
    Parameters
    n/a — dataset
    Context
    unknown
    Languages
    Danish
    Licence
    unknown
    Release
    unknown

Profile note

Denmark's applied-research bridge between universities and industry, owned by the Aarhus University Research Foundation. Maintains DaNLP (the standard Danish NLP resource collection) and DaNE, and is a core partner in Danish Foundation Models. Its most interesting current output is a large family of small per-language hallucination-detection classifiers built on mmBERT — 70 models on HuggingFace, actively updated as of Aug 2026 — which is a genuinely useful RAG-evaluation asset for European languages. No commercial API.

Record sources

  1. https://huggingface.co/alexandrainst
  2. https://github.com/alexandrainst/danlp
  3. https://huggingface.co/datasets/alexandrainst/dane
  4. https://oecd.ai/en/dashboards/policy-initiatives/secure-platform-for-developing-transparent-danish-language-m…
  5. https://alexandra.dk/about-the-alexandra-institute/