---
license: gemma
datasets:
- NbAiLab/aurora-sft-2512-filtered
language:
- 'no'
- nb
- nn
base_model: NbAiLab/borealis-4b-instruct-preview
pipeline_tag: text-generation
library_name: mlx
tags:
- conversational
- instruct
- experimental
- mlx
---

# Borealis 4B Instruct MLX (Preview)

Release: Dec 22nd, 2025.

## Model summary
**NbAiLab/borealis-4b-instruct-preview-mlx** is a MLX 8bit quantized version of a **4B-parameter** instruction-tuned **preview** model intended for early testing and feedback. It is an **experiment** and should be treated as pre-release quality.

The original model is [NbAiLab/borealis-4b-instruct-preview](https://huggingface.co/NbAiLab/borealis-4b-instruct-preview).

| Model | Bits | Format |
|---|---:|---|
| [NbAiLab/borealis-4b-instruct-preview](https://huggingface.co/NbAiLab/borealis-4b-instruct-preview) | BF16 | Transformers (safetensors) |
| [NbAiLab/borealis-4b-instruct-preview-gguf](https://huggingface.co/NbAiLab/borealis-4b-instruct-preview-gguf) | 8 | GGUF (`q8_0`) |
| [NbAiLab/borealis-4b-instruct-preview-gguf](https://huggingface.co/NbAiLab/borealis-4b-instruct-preview-gguf) | 16 | GGUF (`f16`) |
| [NbAiLab/borealis-4b-instruct-preview-gguf](https://huggingface.co/NbAiLab/borealis-4b-instruct-preview-gguf) | BF16 | GGUF (`bf16`) |
| [NbAiLab/borealis-4b-instruct-preview-mlx](https://huggingface.co/NbAiLab/borealis-4b-instruct-preview-mlx) | 32 | MLX |
| [NbAiLab/borealis-4b-instruct-preview-mlx-8bits](https://huggingface.co/NbAiLab/borealis-4b-instruct-preview-mlx-8bits) | 8 | MLX (quantized) |


This model [NbAiLab/borealis-4b-instruct-preview-mlx-8bits](https://huggingface.co/NbAiLab/borealis-4b-instruct-preview-mlx-8bits) was
converted to MLX format from [NbAiLab/borealis-4b-instruct-preview](https://huggingface.co/NbAiLab/borealis-4b-instruct-preview)
using mlx-lm version **0.29.1**.

## Use with mlx

```bash
pip install mlx-lm
```

```python
from mlx_lm import load, generate

model, tokenizer = load("NbAiLab/borealis-4b-instruct-preview-mlx-8bits")

prompt = "hei :)"

if tokenizer.chat_template is not None:
    messages = [{"role": "user", "content": prompt}]
    prompt = tokenizer.apply_chat_template(
        messages, add_generation_prompt=True
    )

response = generate(model, tokenizer, prompt=prompt, verbose=True)
```