
Independent company · Germany
ellamind
A German AI company, based in Bremen and Berlin, that builds evaluation tools and open-source infrastructure for language models.
Germany
- Type
- Independent company
- Based in
- Germany
- Key papers
- 1
- Featured projects
- 8
About
ellamind GmbH works on evaluation, data and environments for AI systems and helps regulated industries such as health insurance, banking and public administration bring them into production. Its products include elluminate, an agentic evaluation platform, and an accessibility testing agent.
The company open-sources base-eval, curated lm-evaluation-harness task configurations for English and German base models, and inference-hive, distributed LLM inference for SLURM clusters, both under Apache-2.0. It also publishes propella-1, a family of small multilingual annotation models, and sui-1, a 24B-parameter summarization model with inline citations, on Hugging Face.
Mission and ambitionsellamind focuses on making AI systems measurable and dependable for organisations in regulated sectors, and shares part of its tooling as open source.
Known for
- 01
Open evaluation tooling for German models
- 02
inference-hive for SLURM clusters
- 03
propella-1 annotation models
Open source
Automatic hyperparameter optimisation of RAG pipelines (8 stars)
Base model evaluation (8 stars)
Evaluation toolkit (7 stars)
Toolkit for distributed LLM inference on SLURM clusters (4 stars)
4B document annotation model, ellamind's most downloaded model.
0.6B document annotation model.
- ellamind/arc-multilingualdataset
Multilingual ARC benchmark.
- ellamind/base-evalspace
Base model evaluation space.
Research papers
Notable releases
Openness
Code is mostly MIT or Apache-2.0; models and datasets are public on Hugging Face.
Contact and community
The company site has a contact page and a demo booking form.
Sources
Compiled from the lab's public pages, Hugging Face and GitHub. Last checked 2026-11-17. Something wrong or out of date? Tell us on Discord.



