---
title: NuExtract3-mlx-4bits
canonical_url: "https://www.modelscope.ai/models/numind/NuExtract3-mlx-4bits"
md_url: "https://www.modelscope.ai/models/numind/NuExtract3-mlx-4bits.md"
repository: numind/NuExtract3-mlx-4bits
last_updated: 2026-05-22
model_type:
  - qwen3_5
architectures:
  - Qwen3_5ForConditionalGeneration
base_model:
  - numind/NuExtract3
base_model_relation: quantized
parameters: 991.5M
tensor_type:
  - U32
  - BF16
library_name:
  - mlx
  - safetensors
  - pytorch
frameworks:
  - pytorch
downloads: 167
stars: 0
tags:
  - mlx
  - mlx-vlm
  - quantized
---

# NuExtract3-mlx-4bits

> NuExtract3-mlx-4bits - An open-source model by numind on ModelScope. numind/NuExtract3-mlx-4bits

numind/NuExtract3-mlx-4bits is a 991.5M-parameter machine learning model on ModelScope. derived from numind/NuExtract3.

- **Repository**: numind/NuExtract3-mlx-4bits
- **Parameters**: 991.5M
- **Base model**: numind/NuExtract3
- **Tags**: mlx, mlx-vlm, quantized
- **Downloads**: 167
- **Stars**: 0
- **Last updated**: 2026-05-22

Source: https://www.modelscope.ai/models/numind/NuExtract3-mlx-4bits

---

# numind/NuExtract3-mlx-4bits

MLX quantized version of `numind/NuExtract3`.

## Quantization

- mode: `affine`
- group_size: `64`
- bits: `4`
- dtype: `source/default`
- quant_predicate: `none`

## Usage

```bash
pip install -U mlx-vlm
mlx_vlm.generate --model numind/NuExtract3-mlx-4bits --prompt "your prompt"
```