| Reference number |
2026-00355 |
| Coordinator |
Lindholmen Science Park AB - AI Sweden
|
| Funding from Vinnova |
SEK 15 652 406 |
| Project duration |
February 2026 - January 2028
|
| Status |
Ongoing |
Purpose and goal
The aim of the project is to strengthen Sweden´s and Europe´s digital sovereignty through the development of fully open, multilingual and high-performance large-scale language models (OpenEuroLLM). AI Swedens´s participation ensures that Swedish linguistic, cultural and application needs are covered and integrated into the models from the ground up.
Expected effects and result
The project will deliver open and license-free European language models of various sizes. For AI Sweden this will result in methods and frameworks for model refinement and extended context length. The effect will be a democratized access to safe AI for the Swedish and European sector, where the Swedish language is well represented without dependence of closed commercial actors.
Planned approach and implementation
OpenEuroLLM is implemented in a consortium of 20 partners. AI Sweden leads the work on the model´s post-training (T4.6), which includes instruction and reasoning training. We are also responsible for data packaging (T3.5) and contribute to a customized tokenizer. The training is performed on European supercomputers such as LUMI, MareNostrum5, Jupiter and Leonardo, and the quality is continuously evaluated.
External links
The project description has been provided by the project members themselves and the text has not been looked at by our editors.
