borealis-27B-instruct-preview: benchmark results
NB-AiLab at the National Library of Norway publishes Borealis 27B Instruct Preview, a Norwegian instruction fine-tune of google/gemma-3-27b-it. This is the experimental Transformers checkpoint fine-tuned on textual instructions, distinct from its GGUF and MLX conversion releases. Provider: National Library of Norway. Released 2026-01-21. Access: Open.
Aggregate Rating 1617 ± 34, rank #864 of 10916 rated models, from 47 benchmark results.
Strongest benchmark results
| Benchmark | Score | Metric | Percentile |
|---|---|---|---|
| EuroEval Faroese NLU - FoQA | 78.65 | Character F1 (%) | 97.2 |
| EuroEval Danish NLU - Angry Tweets | 59.53 | Matthews correlation coefficient (%) | 97 |
| EuroEval Swedish NLU - Swerec | 80.08 | Matthews correlation coefficient (%) | 95.6 |
| EuroEval Norwegian NLU - ScaLA NN | 60.45 | Matthews correlation coefficient (%) | 93.2 |
| EuroEval Faroese | 59.3 | Average Score (%) | 92.6 |
| EuroEval Icelandic NLU - Hotter and Colder Sentiment | 54.51 | Matthews correlation coefficient (%) | 90.8 |
| EuroEval Norwegian Knowledge - Idioms NO | 43.47 | MCC (x100) | 89.2 |
| EuroEval Faroese NLU - FoSent | 62.01 | Matthews correlation coefficient (%) | 88.7 |
| EuroEval Norwegian NLU - Norec | 59.11 | Matthews correlation coefficient (%) | 88.2 |
| EuroEval Icelandic NLU - NQII | 56.09 | Character F1 (%) | 87.7 |
| EuroEval Icelandic Common Sense Reasoning | 58.22 | Common Sense Reasoning Average Score (%) | 87.3 |
| EuroEval Norwegian NLU | 64.88 | NLU Average Score (%) | 87.1 |
Interactive version: theaggregate.ai/model?slug=borealis-27b-instruct-preview · How It Works · Data refreshed daily, snapshot 2026-11-08.