NbAiLab/borealis2-e2b-preview

VISIONPricing:Input $0.4 / Cached $0.08 / Output $0.8Concurrent Unit Cost:1Model Size:5.1BQuant:BF16Context Size:32kTool Calling:SupportedPublished:Sep 28, 2026License:otherArchitecture:Transformer Featherless Exclusive Cold

NbAiLab/borealis2-e2b-preview is a 5.1 billion parameter instruction-tuned preview model developed by NbAiLab, based on NbAiLab/nb-gpt-gemma4-e2b-instruct-aurora-2604-pre. It is continuously pre-trained on the 47 billion word 'aurora' dataset with a knowledge cutoff of January 2025, and fine-tuned on textual instructions. This model is designed for Norwegian-centric assistant tasks, including drafting, summarization, Q&A, and light reasoning, with a focus on Norwegian (Bokmål, Nynorsk) and English language coverage.

Loading preview...

Borealis 2 E2B (Preview) Overview

NbAiLab/borealis2-e2b-preview is a 5.1 billion parameter instruction-tuned model from NbAiLab, serving as an early preview for testing and feedback. It is built upon the NbAiLab/nb-gpt-gemma4-e2b-instruct-aurora-2604-pre base model and has undergone continuous pre-training on the extensive aurora dataset, which comprises 47 billion whitespace-separated words with a knowledge cutoff of January 1st, 2025. The model was then supervised fine-tuned for three epochs using the NbAiLab/aurora-sft-2606 dataset, with SFT loss calculated exclusively on assistant responses.

Key Capabilities & Characteristics

  • Norwegian Language Focus: Optimized for Norwegian (Bokmål and Nynorsk) and English, leveraging a diverse dataset including hplt-3.0, fineweb-2, nb-newspapers-mbl, and finepdfs with significant Norwegian content.
  • Instruction-Tuned: Designed to follow textual instructions for various assistant-style tasks.
  • Preview Quality: Intended for early evaluation, offering insights into its behavior, language coverage, and overall quality.

Intended Use Cases

  • Norwegian-centric Assistant Tasks: Ideal for drafting, summarization, question-answering, and light reasoning in Norwegian.
  • Norwegian Writing Assessment: Useful for evaluating Norwegian writing style and quality.
  • Early Evaluation: Suitable for assessing model behavior and language capabilities in a preview context.

Important Considerations

This model is a preview experiment and has not been fully safety-aligned. Users should be aware that it may produce harmful, biased, or inappropriate content. It is not recommended for safety-critical or high-stakes applications without additional safety mitigations. The license includes restrictions against recreating training data or using the model for services providing access to licensed press publications.

Finetunes
1 models