---
library_name: mlx
tags:
- mlx
- oq
- quantized
---

> [!IMPORTANT]
> This quantization was uploaded on **2026-07-01** and replaces a previous version.
> If you downloaded this model before this date, please re-download for the updated weights.

# NuMarkdown-8B-Thinking-oQ8

This model was quantized using [oQ](https://github.com/jundot/omlx) (oMLX v0.4.5.dev1) mixed-precision quantization.

## Quantization details

- **Model type**: qwen2_5_vl
- **Bits**: 8
- **Group size**: 64
- **Format**: MLX safetensors