NbAiLab/borealis-open-1b

TEXT GENERATIONConcurrency Cost:1Model Size:1BQuant:BF16Ctx Length:32kPublished:May 20, 2026License:gemmaArchitecture:Transformer0.0K Cold

NbAiLab/borealis-open-1b is a 1 billion parameter instruction-tuned language model developed by the National Library of Norway, based on the Gemma 3 architecture. This model is specifically fine-tuned for Norwegian-centric instruction following, excelling in tasks such as assistant-style interactions, writing, summarization, and question answering in Norwegian. It is an open release variant, trained on an open supervised fine-tuning dataset focused on Norwegian language tasks.

Loading preview...

Borealis Open 1B: Norwegian-Centric Instruction Model

NbAiLab/borealis-open-1b is a 1 billion parameter instruction-tuned model from the National Library of Norway (NbAiLab), part of the Borealis family. Built upon the Gemma 3 architecture (specifically google/gemma-3-1b-it), this model is specialized for Norwegian language tasks.

Key Capabilities & Features

  • Norwegian Language Focus: Instruction-tuned for Norwegian-centric assistant behavior, writing, summarization, and question answering.
  • Open Release: This model is an "open" variant, meaning it does not include material from specific rights-holder agreements, unlike the full Borealis models.
  • Safety Alignment: Utilizes "prompt baking" for safety alignment, distilling system prompt behavior directly into model weights to balance quality, usefulness, and safer outputs.
  • Context Length: Supports a context length of 32768 tokens.

Intended Use Cases

  • Norwegian Assistant Tasks: Ideal for drafting, summarization, Q&A, and light reasoning in Norwegian.
  • Writing Assessment: Useful for assessing and improving Norwegian writing style and quality.
  • Language Coverage Evaluation: Suitable for evaluating behavior and language coverage across Norwegian, Bokmål, and Nynorsk.

Limitations

  • May produce incorrect or biased information.
  • Performance outside Norwegian and English use cases is not fully characterized.